FEDERAL UNIVERSITY OF SERGIPE

OFFICE OF THE DEAN OF GRADUATE STUDIES AND RESEARCH

GRADUATE PROGRAM IN EDUCATION
PhD IN EDUCATION

ALLESSANDRA ELISABETH DOS SANTOS

GENERATIVE ARTIFICIAL INTELLIGENCE AND ITS IMPACT ON WRITING

SÃO CRISTÓVÃO-SE-BRAZIL
2024
 ALLESSANDRA ELISABETH DOS SANTOS

GENERATIVE ARTIFICIAL INTELLIGENCE AND ITS IMPACT ON WRITING

Doctoral dissertation presented to the Graduate

Program in Education (PPGED) as a requirement
serving the application phase for the Ph.D. in
Education at the Federal University of Sergipe.
Area: Education
Field of study: Technologies, Languages, and
Education

Advisor: Prof. Dr. Paulo Roberto Boa Sorte Silva

Co-advisor: Prof. Dr. Neiva da Silva Rego Ravagnoli

SÃO CRISTÓVÃO-SE-BRAZIL
2024
FICHA CATALOGRÁFICA ELABORADA PELA BIBLIOTECA CENTRAL
UNIVERSIDADE FEDERAL DE SERGIPE

Santos, Allessandra Elisabeth dos

S237g Generative Artificial Intelligence and its impact on writing /
Allessandra Elisabeth dos Santos ; advisor Paulo Roberto Boa
Sorte Silva. – São Cristóvão, SE, 2024.
156 f. : il

Thesis (doctorate in Education) – Federal University of Sergipe,

2024.

1. Education. 2. Artificial intelligence – Effect of [...] on. 3.

Language and education. 4. ChatGPT. 5. Writing. I. Silva, Paulo
Roberto Boa Sorte, advisor. II. Title.

CDU 37:004.8:81
 In loving memory of Dinda Marisa,
Vovó Carmosa,
and my sweet Paris,
whose spirit and love continue to inspire me.
 SPECIAL ACKNOWLEDGMENTS

To my esteemed advisor, Prof. Dr. Paulo Boa

Sorte Silva, thank you for the countless lessons
and the unwavering support throughout my
research journey. Your belief in my abilities,
even before I could see them myself, has been the
true hallmark of your role as an educator. This
trust has been instrumental in shaping my path as
a researcher. You embody more than a
mentorship role; you are a profound source of
inspiration and knowledge, bridging both
academic and personal spheres. My deepest
appreciation and respect are forever yours.
 SPECIAL ACKNOWLEDGMENTS

I am immensely grateful for the guidance and

support of my co-advisor, Prof. Dr. Neiva
Cristina da Silva Rego Ravagnoli, a
knowledgeable researcher in Complex Thinking,
who has skillfully guided me along the learning
path of this epistemology. Your gentle, friendly,
and humorous approach has always made me feel
welcomed, greatly aiding my learning process. It
has been a privilege to have you shape my
formation as a researcher, starting from my
Master’s degree and continuing through my
doctoral journey. My deepest appreciation and
respect are forever yours.
 ACKNOWLEDGMENTS

I extend my sincere gratitude to Prof. Dr. Luciana C. de Oliveira, my PDSE Supervisor

from Virginia Commonwealth University (VCU), for her instrumental role in my development
as a researcher. Her willingness to integrate me into collaborative writing projects with
members of her research group has opened numerous doors and provided me with an enriching
learning experience. I cannot thank her enough for the opportunities to co-write with her and
her team, which have greatly enhanced my understanding and skills. Her support and
mentorship have been invaluable, and for this, I am profoundly thankful.
I am deeply grateful to Prof. Dr. Simone de Lucena Ferreira for her valuable
contributions to my dissertation at every stage. I also thank her for her constant presence and
unwavering support throughout my academic journey, starting from my master’s degree. Her
guidance has enhanced my work and has been crucial in shaping my development as a
researcher.
I am deeply appreciative of Prof. Dr. David Kohan Margazao for his significant role
during my qualifying exam. His thought-provoking questions and challenges have not only
deepened my understanding but also encouraged me to pursue a more insightful research path.
His support has been instrumental in my academic growth, and I am immensely grateful for his
guidance.
I am profoundly thankful to Prof. Dr. José Paulo Gomes Brazão for his insightful
contributions during my qualifying exam. The questions and challenges he presented were
crucial in guiding me toward a more fulfilling research journey. His guidance has been
indispensable in defining the direction of my work. Additionally, his expert feedback has
greatly enriched my research, enhancing both its depth and clarity.
I am also grateful to the friends I have made during this journey, especially my
roommates Chardé Reid and Lily, who made my stay in Richmond a memorable experience,
and Ben Olsoe, who welcomed me during my PhD Sandwich Program at VCU and who became
a valued friend. Thank you for your support and friendship.
I would like to acknowledge the friends with whom I collaborated on writing projects:
Larisa Olesova, Cristiane Vicentini, Karen Terrell, Jia Gui, Tara Willging, Joy Beatty, and
Destini Braxton. Their insights and teamwork were instrumental in enhancing the quality of my
writing.
In addition, I want to extend my gratitude to the members of the TECLA research group,
whose contributions during our meetings significantly improved my research. Their collective
expertise and collaborative approach played a significant role in refining my research and
advancing this dissertation.
Special thanks to my dear friend/sister, Laila Gardênia, for her exceptional technical
assistance in refining the layout of my Prompt Creation Reference Chart and some images in
my dissertation, as well as for her emotional support throughout this challenging journey. I also
extend my gratitude to Nadege Siqueira, Bruna Oliveira, and Jefferson Santos for their
dedicated and constant support.
I would like to extend my heartfelt gratitude to all my beloved family members,
especially my parents, Adalberto Antônio dos Santos and Josefa Nelma dos Santos, for the love
and dedication they have provided me throughout my years of study. I am also deeply thankful
to my uncle, who has been a father figure, João Batista dos Santos, and to my youngest brother,
Allisson Mário dos Santos, for their encouragement and loving presence in my life.
I wish to thank my partner, Mirco Buiano, for his support and dedication over these four
years. Despite the challenges of living in different countries, his companionship and
encouragement throughout my writing process have been invaluable. His love and commitment
have been essential in helping me navigate this challenging journey.
Finally, I would like to thank the Coordination for the Improvement of Higher Education
Personnel (CAPES) for the financial support provided through the scholarship, which was
essential for the development of this research.
 ABSTRACT

AI: word of 2023. The abbreviation of Artificial Intelligence (AI) was named the word of the
year by Collins Dictionary, reflecting its growing prominence in everyday life. Although AI
has long permeated various aspects of human activity, our society only began recognizing its
ubiquitous presence after ChatGPT was released on November 30, 2022. ChatGPT, an AI
chatbot that produces human-like responses, exemplifies Generative Artificial Intelligence
(GenAI) technologies now widely accessible, provoking discussions and raising ethical
concerns across social, political, economic, and academic contexts. In education, technologies
undeniably reshaped writing, a social practice grounded in evolving literacy perspectives.
Recognizing these transformations, the central question of this study is: How does the
integration of GenAI technologies, like Large Language Models (LLMs), redefine the concept
of writing in educational contexts? To address this question, this research aims to apprehend
the concept of the phenomenon of writing impacted by the adoption of GenAI in societal
contexts in the field of education. Additionally, the specific objectives are to examine the
approach to writing conveyed on internet-based platforms, particularly after the launch of
ChatGPT, and to explore educational, pedagogical, and ethical issues related to the impact of
AI-text generators, especially ChatGPT, raised on these platforms. Considering that the study
of the phenomenon of writing cannot be done in isolation from our society, knowledge, and
culture, it is crucial to adopt an epistemological and methodological approach that rejects
oversimplifying realities. Guided by Complex Thinking (Morin, 2015), this qualitative,
documentary, and bibliographic study analyzed 60 documents — including digital newspaper
articles, social media posts, YouTube videos, podcasts, and websites — using Analytical
Entwining integrated with GenAI. Findings revealed eight theoretical constructs. The key terms
— revolution, integrity, and transformation — were extracted from the first three constructs
and address the first specific objective of this research. While the key terms — enhancement,
support, quality, authenticity, equity, and integration — extracted from five theoretical
constructs, address the second objective. By situating writing as an evolving social practice that
interacts reciprocally with GenAI, the study suggests that GenAI broadens writing as a social
technology, reinforcing the need for adaptive strategies that ensure AI augments rather than
replaces human agency. The integration of GenAI in writing, analyzed through the principles
of Complex Thinking, reveals a complex interplay between technology and human agency, with
both elements in constant adaptation, influencing each other systemically, holographically,
recursively, and dialogically, ensuring that the writing process remains dynamic and grounded
in humanistic values. Recognizing writing as part of this evolving system, this research
introduces the concept of writing as an ecosystem, in which technology and human elements
are interdependent and mutually enhancing, contributing to a more collaborative and adaptive
approach to writing instruction.

Keywords: generative artificial intelligence; large language models; ChatGPT; education;

writing.
 RESUMO

IA: palavra de 2023. A abreviatura de Inteligência Artificial (IA) foi nomeada a palavra do ano
pelo Dicionário Collins, refletindo sua crescente proeminência na vida cotidiana. Embora a IA
já permeasse vários aspectos da atividade humana, nossa sociedade apenas passou a reconhecer
a sua presença ubíqua após o lançamento do ChatGPT, em 30 de novembro de 2022. O
ChatGPT, um chatbot de IA que gera respostas semelhantes às de humanos, exemplifica as
tecnologias de Inteligência Artificial Generativa (IAG) atualmente amplamente acessíveis,
provocando discussões e suscitando preocupações éticas em contextos sociais, políticos,
econômicos e acadêmicos. No campo da educação, as tecnologias inegavelmente remodelaram
a escrita, entendida como uma prática social fundamentada em perspectivas de letramentos em
constante evolução. Reconhecendo essas transformações, a questão central deste estudo é:
Como a integração de tecnologias de IAG, como Modelos de Linguagem de Grande Escala
(LLMs), redefine a concepção de escrita em contextos educacionais? Para responder a essa
pergunta, esta pesquisa objetiva apreender a concepção do fenômeno escrita impactado pela
adoção da IAG em contextos sociais na área da educação. Além disso, propõe-se como
objetivos específicos examinar a abordagem da escrita veiculada em plataformas digitais,
sobretudo após o lançamento do ChatGPT, e explorar questões educacionais, pedagógicas e
éticas relacionadas ao impacto de geradores de texto em IA — especialmente o ChatGPT —
levantadas nessas plataformas. Considerando que o estudo do fenômeno escrita não pode se
dissociar de nossa sociedade, conhecimento e cultura, torna-se fundamental adotar uma
perspectiva epistemológica e metodológica que rejeite a simplificação excessiva das realidades.
Orientado pelo Pensamento Complexo (Morin, 2015), este estudo qualitativo, de caráter
documental e bibliográfico, analisou 60 documentos — incluindo artigos de jornais digitais,
publicações em redes sociais, vídeos no YouTube, podcasts e sites — utilizando o Entrelace
Analítico integrado à IAG. Os resultados evidenciaram oito construtos teóricos. Os termos-
chave — revolução, integridade e transformação — foram extraídos dos três primeiros
construtos e atendem ao primeiro objetivo específico desta pesquisa. Já os termos-chave —
aprimoramento, suporte, qualidade, autenticidade, equidade e integração — extraídos de cinco
construtos teóricos, respondem ao segundo objetivo. Ao situar a escrita como uma prática social
em constante evolução e em interação recíproca com a IAG, o estudo sugere que a IAG amplia
a escrita como tecnologia social, reforçando a necessidade de estratégias adaptativas que
garantam a complementação, em vez da substituição, da agência humana. A integração da IAG
à escrita, analisada sob os princípios do Pensamento Complexo, revela uma interação complexa
entre tecnologia e agência humana, com ambos os elementos em contínua adaptação,
influenciando-se mutuamente de maneira sistêmica, holográfica, recursiva e dialógica. Isso
assegura que o processo de escrita permaneça dinâmico e ancorado em valores humanísticos.
Reconhecendo a escrita como parte desse sistema em evolução, esta pesquisa introduz a
concepção de escrita como ecossistema, em que aspectos tecnológicos e humanos são
interdependentes e se fortalecem mutuamente, contribuindo para uma abordagem de ensino da
escrita mais colaborativa e adaptativa.

Palavras-chave: inteligência artificial generativa; modelos de linguagem de grande escala;

ChatGPT; educação; escrita.
 LIST OF FIGURES

Figure 1 – How worried should you be about AI disrupting elections? ................................... 22

Figure 2 – ChatGPT and the 2024 US election ........................................................................ 23
Figure 3 – Imagined arrest images of Donald Trump created using Midjourney v5 ............... 24
Figure 4 – ChatGPT: 1 million active users ............................................................................. 25
Figure 5 – Article: What is in store with the GPT-3 algorithm? .............................................. 27
Figure 6 – First published article generated by GPT-3 large language model ......................... 27
Figure 7 – ChatGPT is banned in Italy ..................................................................................... 28
Figure 8 – ChatGPT is banned in NYC public schools ............................................................ 29
Figure 9 – Teachers’ reactions towards ChatGPT .................................................................... 30
Figure 10 – Research concept map ........................................................................................... 36
Figure 11 – Scientific publications and presentations .............................................................. 39
Figure 12 – My research in light of Complex Thinking ........................................................... 47
Figure 13 – Complex Thinking: research guiding questions .................................................... 49
Figure 14 – GPT-3 in journalism.............................................................................................. 50
Figure 15 – AI: Fantástico News report................................................................................... 50
Figure 16 – AI text generator selection criteria ........................................................................ 52
Figure 17 – Complex Thinking: my research guiding questions.............................................. 54
Figure 18 – My guiding questions ............................................................................................ 55
Figure 19 – Designing my path ................................................................................................ 56
Figure 20 – Analytical Entwining ............................................................................................ 59
Figure 21 – ChatGPT: 100 million active users ....................................................................... 61
Figure 22 – Analytical Entwining Integrated with GenAI ....................................................... 66
Figure 23 – Prompt Creation Reference Chart 1 ...................................................................... 67
Figure 24 – Prompt Creation Reference Chart 2 ...................................................................... 69
Figure 25 – Prompt Creation Reference Chart 3 ...................................................................... 73
Figure 26 – Prompt Creation Reference Chart 4 ...................................................................... 77
Figure 27 – The Journey of OpenAI GPT Models ................................................................... 81
Figure 28 – Methodological Path ............................................................................................. 83
Figure 29 – Theoretical Constructs: specific objective 1 ......................................................... 87
Figure 30 – “Screenshot of the reality I see” 4C ...................................................................... 90
Figure 31 – “Screenshot of the reality I see” 3B ...................................................................... 91
Figure 32 – “Screenshot of the reality I see” 5B ...................................................................... 94
Figure 33 – “Screenshot of the reality I see” 3C ...................................................................... 96
Figure 34 – “Screenshot of the reality I see” 5A ...................................................................... 98
Figure 35 – “Screenshot of the reality I see” 5B .................................................................... 103
Figure 36 – Theoretical Constructs: specific objective 2 ....................................................... 107
Figure 37 – “Screenshot of the reality I see” 1E .................................................................... 110
Figure 38 – “Screenshot of the reality I see” 2E .................................................................... 112
Figure 39 – “Screenshot of the reality I see” 3E .................................................................... 114
Figure 40 – “Screenshot of the reality I see” 1D .................................................................... 117
Figure 41 – “Screenshot of the reality I see” 4D .................................................................... 119
Figure 42 – “Screenshot of the reality I see” 4F..................................................................... 122
Figure 43 – “Screenshot of the reality I see” 5F..................................................................... 126
Figure 44 – “Screenshot of the reality I see” 4D .................................................................... 130
Figure 45 – “Screenshot of the reality I see” 4E .................................................................... 132
Figure 46 – My research journey............................................................................................ 137
Figure 47 – The concept of writing impacted by GenAI ........................................................ 138
 LIST OF TABLES

Table 1 – Data: digital newspaper articles................................................................................ 62

Table 2 – Repeated ideas: initial theme .................................................................................... 62
Table 3 – Repeated ideas: second theme .................................................................................. 63
Table 4 – Theoretical constructs and themes............................................................................ 64
Table 5 – Visualizing Prompt No. 1 ......................................................................................... 68
Table 6 – Visualizing Prompt No. 2 ......................................................................................... 70
Table 7 – Visualizing Output Text Document 4C .................................................................... 71
Table 8 – Visualizing Prompt No. 3 ......................................................................................... 74
Table 9 – Visualizing Output - Prompt No. 3 - Specific Objective 1....................................... 75
Table 10 – Visualizing Prompt No. 4 ....................................................................................... 78
Table 11 – Visualizing Output - Prompt No. 4 - Specific Objective 1..................................... 78
 LIST OF ABBREVIATIONS

AI Artificial Intelligence
AWE Automated Writing Evaluation
CAPES Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
CEP Comitê de Ética em Pesquisa
CESAD Centro de Educação Superior a Distância
DL Deep Learning
GenAI Generative Artificial Intelligence
GLTR Generative Textual Likelihood Ratio
GPT Generative Pre-trained Transformer
ICC International Criminal Court
IOA Intelligent Online Assessment
IoT Internet of Things
LLMs Large Language Models
ML Machine Learning
NLP Natural Language Processing
PMFGN Personalized Multimodal Feedback Generation Network
PDSE Programa de Doutorado Sanduíche no Exterior
PPGED Programa de Pós-Graduação em Educação
TCLE Termo de Consentimento Livre e Esclarecido
TECLA Tecnologias, Educação e Linguística Aplicada
TLMs Transformer Language Models
UFS Universidade Federal de Sergipe
UFSC Universidade Federal de Santa Catarina
UNCISAL Universidade Estadual de Ciências da Saúde de Alagoas
UNINASSAU Centro Universitário Maurício de Nassau
VCU Virginia Commonwealth University
 TABLE OF CONTENTS

1 THE BEGINNING OF MY JOURNEY ........................................................................ 19

2 METHODOLOGICAL PATH ....................................................................................... 43

2.1 Complex Thinking and its core principles .................................................................... 44
2.2 The nature of research .................................................................................................... 48
2.3 The change of course ....................................................................................................... 49
2.4 Redesigning my path ....................................................................................................... 53
2.5 Method of analysis ........................................................................................................... 57

3 APPROACHES TO WRITING IN EDUCATION WITH GENERATIVE AI......... 86

3.1 Writing is revolutionized by AI through enhancement and accessibility .................. 88
3.2 Writing integrity is maintained through AI-driven ethical safeguards ..................... 95
3.3 Writing education is transformed by AI through personalization and global .............
reach................................................................................................................................ 101

4 EDUCATIONAL, PEDAGOGICAL, AND ETHICAL ISSUES OF AI TEXT

GENERATORS ............................................................................................................. 106
4.1 Writing enhances educational practices with AI tools ............................................... 109
4.2 Writing processes supported by AI’s automation and personalization ................... 115
4.3 Writing quality and authenticity in the age of AI ...................................................... 120
4.4 Writing and educational equity with AI tools ............................................................. 124
4.5 Writing integrates AI while preserving human interaction ...................................... 128

5 FINAL REMARKS ....................................................................................................... 136

REFERENCES .............................................................................................................. 141

APPENDIX I – Research credibility............................................................................ 154

APPENDIX II – Scientific publications....................................................................... 156

 19

SECTION 1

THE BEGINNING OF MY JOURNEY1

1
Image generated using OpenAI’s DALL-E image generation tool on November 11, 2023. The prompt used for
generation was: “Generate images that represent a female researcher wearing glasses and with dark hair.
She’s at the beginning of her study path, focusing on education, writing, artificial intelligence, and ChatGPT.
The images depict a woman at a crossroads, surrounded by symbolic elements related to her areas of research”.
 20

1 THE BEGINNING OF MY JOURNEY

“Consciousness of complexity leads us to realize

that we will never escape uncertainty
and that we can never attain total knowledge:
Totality is not the truth” (Morin, 2015, p. 69).

The journey you are about to embark on addresses areas of knowledge intrinsically
linked to the structure of our contemporary societies, which are Artificial Intelligence (AI),
Technologies, Languages, and Education. So, I have decided to present my initial path by
inviting you to reflect upon a quotation by one of the Nobel Peace Prize 2021 winners, Maria
Angelita Ressa, a journalist and author.
“So, who knows what will happen in our May elections? One thing is clear [...] you
cannot have integrity of elections if you don’t have integrity of facts. And that is something all
of us, all democracies around the world face” (Ressa, 2021). Maria Angelita Ressa said this
quotation in an interview on the importance of a free press for PBS NewsHour, a North
American evening television news program. As the Chief Executive Officer (CEO) and co-
founder of Rappler (Rappler, 2021), an online news website based in the Philippines, the 2021
Nobel Peace laureate has faced constant political harassment and arrests by the Duterte
government (Ratcliffe, 2022). Rodrigo Duterte ruled the presidency in the Philippines from
2016 to 2022. He’s known as “The Punisher” due to the high number of extrajudicial killings
in his “war on drugs”, an estimate of up to 30,000 people died according to the International
Criminal Court (ICC). Ressa was forced to post bail various times in order to remain out of
prison. Rappler’s battle for truth and democracy is being charged by the Philippines government
with having committed cyber-libel.

The Norwegian Nobel Committee has decided to award the Nobel Peace Prize
for 2021 to Maria Ressa and Dmitry Muratov for their efforts to safeguard
freedom of expression, which is a precondition for democracy and lasting
peace. Ms. Ressa and Mr. Muratov are receiving the Peace Prize for their
courageous fight for freedom of expression in the Philippines and Russia. At
the same time, they are representatives of all journalists who stand up for this
ideal in a world in which democracy and freedom of the press face
increasingly adverse conditions (Reiss-Andersen, 2021).

The Nobel Peace Committee acknowledges the need for a reform of social media
platforms due to their role in fostering hatred and violence. In Ressa’s speech, she warned about
the manipulation taking place on social media platforms, with algorithms prioritizing lies over
facts. Ressa highlighted the role of technology and social media platforms in aiding the rise of
 21

such regimes and stressed the need for regulation to curb the impunity that social media enjoys.
From Ressa’s perspective, technology acts as the accelerant of disinformation and fake news
since social media has placed a greater emphasis on circulating misinformation and the spread
of hatred over verifiable facts. She calls for a reform of social media platforms since algorithms
of amplification and algorithms of distribution have been programmed by humans with coded
bias, and to be profit-driven.
The development of algorithms that pose risks also arouses the interest of a highly
profitable market, whose financial return and power relations are markedly significant and
could already be proven, as in the Brexit elections in the United Kingdom, and for president of
the United States, in 2016 and Brazil, in 2018. Social media platforms use algorithms to
personalize the content we see by prioritizing information that aligns with our personal beliefs
and interests. The phenomenon is called filter bubbles, a term coined by Pariser (2011). It is
also described as a state of “intellectual isolation” (Srba et al., 2023, p. 4), negatively impacting
democratic discourse. In addition, the creation of informational bubbles in social media is a
consequence of AI and improved algorithms, which leads us to interact with and access content
primarily from people who have consensual lines of reasoning and opinions (Santaella, 2019).
This phenomenon presents the danger that critical issues will get filtered, leaving people
unexposed to views different from their own.
In all these political scenes, we recognize another phenomenon named post-truth
(McIntyre, 2018), in which truth loses its relevant character in the formation of public opinion,
with a preference for information whose basis lies in beliefs and emotions. Although post-truth
(Oxford, 2016), the word of the year 2016, is not a current phenomenon, the ubiquity — the
state of being omnipresent, a modern life concept resulting from hypermobility (Santaella,
2013) — enabled by smartphones in our daily lives, ends up scaling it.
During the Brazilian presidential elections in 2022, the advances in AI technologies
continued to impact the decision-making processes of many voters who were being bombarded
with misinformation (Jeantet, 2022). False narratives include claims about the leftist candidate,
Luiz Inácio Lula da Silva, planning to close churches, among other fake news. Also, far-right
YouTube channels have collected more views compared to leftist channels, when the candidate
and former president Jair Messias Bolsonaro questioned the safety of the electronic voting
machines and the fairness of the electoral process. In response to these false allegations, the
Superior Electoral Court has empowered itself to ban content that undermines the electoral
process, demanding that social media platforms remove such content within an hour or face
fines. Luiz Edson Fachin, Brazilian jurist, lawyer, and president of the Supreme Court during
 22

that period, stated that those serious allegations of fraud had no evidence. Fachin highlighted
“It is about time to stop misinformation and authoritarian populism that puts the 1988
Constitution in jeopardy (RBA, 2022, my translation)”. During a period of elections filled with
misinformation and distrust in institutions, the manifesto entitled Carta à democracia (Letters
to Brazilians) (Paraguassu; Marcello, 2022) was released in order to restore faith in the
democratic process, emphasize the relevance of unbiased information as well as encourage
unity among Brazilian citizens.
Again, the United States faces all these concerns regarding disinformation campaigns
for the 2024 presidential elections. There is major constant worry expressed in the media due
to the rapid spread of misinformation made possible by the advancement of AI. The discussions
clearly show the fear that these campaigns may have implications for democracy (see Figure
1).

Figure 1 – How worried should you be about AI disrupting elections?

Source: The Economist (2023).

According to Vallarino (2023), this news article’s author, disinformation is not new in
our society, and is not only created by AI. However, he predicts a new scenario with the
advancements in generative artificial intelligence (GenAI), because of “[…] models that can
spit out sophisticated essays and create realistic images from text prompts—make synthetic
 23

propaganda possible”. The author highlights that there is considerable concern and fear, not
only for the 2024 US presidential elections but also for future campaigns in other countries such
as India, Indonesia, Mexico, and Taiwan.
The Year 2024 stands as a monumental milestone in the global democratic landscape,
marked as “The Ultimate Election Year”. This distinction arises not merely from the frequency
of elections but from their extensive reach and potential impact. In this year, an unprecedented
number of nations—specifically, 64 countries plus the European Union—will engage in
national elections. These elections encapsulate nearly half of the world’s population, about
49%, underscoring their significant democratic engagement. The outcomes of these elections
are poised to shape the geopolitical and social trajectories of these nations for years to come,
highlighting 2024 as a critical year in fortifying democratic processes and principles worldwide
(Ewe, 2023).
In an interview on GZERO World, the attention is also directed to the fact that 2024 is
the first US presidential election in the age of GenAI (Bremmer; Stelter; Hemmer, 2023). There
are concerns about the potential for AI-text and AI-image generator platforms, for instance
ChatGPT (developed by OpenAI) and Midjourney (developed by Midjourney, Inc.), capable of
spreading misinformation and disinformation more rapidly and easily than ever (see Figure 2).

Figure 2 – ChatGPT and the 2024 US election

Source: GZero World (2023).

These media experts say “[…] the stakes are incredibly high for truth and democracy”
(Bremmer; Stelter; Hemmer, 2023) based on the US previous elections. They mention as
 24

consistent examples of these issues the spread of disinformation in social media platforms in
2016, and the conspiracy theories and baseless claims regarding voter fraud in the 2020 election.
They agree on this, due to the amount of “information pollution in the atmosphere” by also
referring to the AI-faked images of Donald Trump’s imagined arrest (Higgins, 2023) (see
Figure 3).
Figure 3 – Imagined arrest images of Donald Trump created using Midjourney v5

Source: Higgins (2023).

In preparation for the 2024 elections in democratic processes, OpenAI has joined the
Coalition for Content Provenance and Authenticity (C2PA)2 to support digital content
certification standards, helping authenticate AI-produced images and videos. The organization
has upgraded its systems to connect users with authoritative election information, including
integrating ChatGPT with official election websites (OpenAI, 2024).
OpenAI’s concern regarding the elections involves the responsible deployment and use
of its GenAI technologies, including the most recent AI text-to-video generation model. Sora is
still a research output that will be launched this year. However, OpenAI has informed that it
will not release any technology, including Sora, that could adversely affect worldwide elections
or propagate misinformation and biases in 2024. Given the potential for these technologies to
be misused, including creating misleading or deceptive content that could affect election
outcomes, OpenAI has implemented measures to safeguard against such risks. This includes
refining usage policies for AI tools, enhancing the transparency of AI-generated content, and
promoting access to reliable voting information. Additionally, OpenAI supports legislative
efforts, such as the “Protect Elections from Deceptive AI Act” which aims to prohibit the

2
Available at: [Link]
 25

distribution of deceptive AI-generated media in political advertising, thereby ensuring the

integrity of electoral processes (OpenAI, 2024).
In these opening pages, I contextualize the emergence of AI and GenAI technologies
within political, economic, and social spheres, highlighting their potential to raise serious
ethical concerns—a situation that has dramatically intensified following the launch of
ChatGPT. This AI-powered large language model was developed by OpenAI, a North
American company established in 2015. Initially founded as a non-profit AI research institution
(Open, 2023), OpenAI has articulated its mission to promote and develop user-friendly AI that
benefits humanity as a whole. The organization pledges to “freely collaborate” with other
institutions and researchers by making its patents and research openly accessible. Supported by
commitments exceeding $1 billion, the company plans a cautious allocation of these funds,
particularly in the initial years. The founders are partly motivated by the existential risks
associated with GenAI.
On November 30, 2022, OpenAI made ChatGPT accessible to the public. This GPT-
3.5-algorithm-based chatbot became a topic of discussion in the mass media because of its
capabilities of generating human-like content. When provided a prompt, this AI-text generator
can perform a wide range of tasks: text generation and completion; language translation;
information summarization; question answering; conversational AI; writing assistance in
composing emails, essays, reports or creative writing pieces like stories, poems, jokes;
educational support providing explanations on academic topics, among others. The list of
possibilities has become endless and its responses can be indistinguishable from human
responses. Its continuing massive popularity achieved one million users in only five days after
its release (see Figure 4).

Figure 4 – ChatGPT: 1 million active users

Source: Yahoo Finance (2022).

 26

Although OpenAI informs ChatGPT has limitations, it is undeniable that this AI chatbot
has achieved a major milestone compared to other platforms. Social media posts, digital
newspaper news, YouTube videos, and podcasts started reporting reactions across the United
States, Brazil, as well as other countries. Even though I am Brazilian, at the time ChatGPT was
rapidly gaining popularity, I was living in the United States.
From September 2022 to May 2023, I was doing my Doctoral “Sandwich” program,
Programa Institutional de Doutorado Sanduíche no Exterior (PDSE), at Virginia
Commonwealth University (VCU) in Richmond, Virginia, as a CAPES 3 scholarship holder.
When I first arrived at VCU, in fact, since I had started working on my doctoral project in 2020,
I used to be told: “I don’t quite understand what your research is about, but it sounds
interesting”. This was an extremely common opinion because, although GPT-3-algorithm-
based text generators already existed, they did not match the immense popularity that ChatGPT
has achieved.
Between mid-2020, marked by the launch of GPT-3, and late November 2022, there
was a significant number of AI-writing platforms available on the market: ShortlyAI, JasperAI,
Simplified, Writesonic, Wordplay, [Link], to name a few. There was a concurrent
emergence of media and academic discourses. It is crucial to mention the article Artificial
intelligence in academic writing: what is in store with the GPT-3 algorithm? (Boa Sorte et al.,
2021). This article centered its discussion on concerns about the impacts on writing in terms of
ethics, creativity, and intellectual property. This scientific paper (see Figure 5) was published
by my advisor Prof. Dr. Paulo Boa Sorte, and co-authored by me and our research group
members: Prof. Dr. Mário Freitas, Jefferson Santos, and Jamile Dias, on October 8, 2021 and it
became a reference after the launch of ChatGPT.

3
CAPES, which stands for Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, is the publications
portal of the Coordination for the Improvement of Higher Education Personnel whose mission is: to promote
graduate programs in Brazil through online access to scientific information.
 27

Figure 5 – Article: What is in store with the GPT-3 algorithm?

Source: Boa Sorte et al. (2021).

In terms of media discourse, the news article4 published by The Guardian on September
8, 2020, was the first piece generated by a GPT-3 algorithm platform reported in the media (see
Figure 6). This article raised a multitude of ethical concerns due to the content it generated and
questions regarding the potential disappearance of professions, especially in the fields of
journalism and cinema.

Figure 6 – First published article generated by GPT-3 large language model

Source: The Guardian (2020).

The amount of media coverage on these issues was relatively limited in comparison to
the extensive media discourse since the advent of ChatGPT in late 2022. I recall having frequent
conversations with my advisor about the reactions I observed when the topic started being
addressed in lectures in various departments at VCU and other US universities. In addition, it
was when the number of news and posts I gathered about my theme of research began to
increase significantly. The vast majority of those responses follows the narrative of the

4
Available at: [Link]
 28

technological phenomenon along human history (Lemos, 2013), in which we find ourselves
confronted with a new technology and our overall tendency is to respond with either
enchantment or fear. In the case of ChatGPT, many of the first reactions resulted in extreme,
negative, and critical attitudes (see Figure 7).

Figure 7 – ChatGPT is banned in Italy

Source: Wired (2023).

At the end of March 2023, Italy decided to temporarily ban ChatGPT (Zorloni, 2023).
According to Garante, the Italian data-protection authority, OpenAI could not legally justify the
collection and storage of users’ personal data to train their algorithms. Italy was the first
Western country to block this chatbot over privacy and ethical concerns. Stanzione (2023), the
president of Garante, described the following issues with the management of personal data by
OpenAI: lack of user information, absence of legal basis for data collection, inaccurate personal
data processing, no verification of users’ ages, risks to users under 13 years old potentially
exposed to inappropriate responses, among others. Despite that, on April 28, 2023, the Garante
removed its limit block of ChatGPT, informing that OpenAI had cooperated in responding to
the Italian authorities.
 29

The absence of regulation for this AI-technology has fueled discussions about its
impacts not only on the social, political, and economic spheres but also in the field of education.
Even before Italy banned access to ChatGPT, the New York City (NYC) Education Department
had decided to block the AI-powered chatbot. In the first week of January 2023, access to
ChatGPT was restricted on devices and networks of NYC public schools (see Figure 8).

Figure 8 – ChatGPT is banned in NYC public schools

Source: Elsen-Rooney (2023).

The reasons for this prohibition were deep worries about potential negative effects on
student learning, and about safety and veracity of the generated content. Furthermore, educators
feared cheating and plagiarism would become more common and this would make writing
assignments obsolete. As the nation’s largest school system in the US, NYC’s decision to
restrict ChatGPT access provokes a ripple effect influencing other school districts. Los Angeles
and Baltimore have agreed on this ban and also blocked access to ChatGPT. It worked as a
strategy so that other school districts had time to evaluate the benefits and risks of the chatbot
(Elsen-Rooney, 2023).
As to teachers’ reactions, they express most concerns on rethinking assessment and
teaching in light of GenAI technologies. The title of the article The college essay is dead clearly
expresses major worries and profound fear towards the impact of these technologies in
education. The image seems to be precisely selected to highlight all these feelings. The author
states concern regarding the historical significance of the essay in humanistic education,
 30

emphasizing its role in teaching research, thinking, and writing skills. Similarly, the article The
end of high school English reacts unfavorably to the possibilities of the incorporation of Large
Language Models (LLMs) in education. The astonishment with the capabilities of ChatGPT
becomes an automatic fearful response (see Figure 9).

Figure 9 – Teachers’ reactions towards ChatGPT

Source: The Atlantic (2022).

A central theme discussed in these three articles is the potential redundancy of

traditional essay writing as a critical metric for learning and intelligence assessment, given
GenAI’s capabilities. The observed reality includes the impact on writing and academic
integrity with concerns about the possible obsolescence of writing assignments, particularly in
the domain of education, and more specifically in high school and humanistic education.
Moreover, the necessity of developing educational policies in view of technological advances
is another reality that could not be ignored as well as the necessary collaboration between
humanities and technology fields. The articles focus on the practical and ethical considerations
of integrating GenAI into a traditionally human-centric domain like education.
In education, the application of AI is present. The sphere of production and
dissemination of texts has been appropriated by AI techniques in its various forms of
representation, i.e., graphic texts, sounds, drawings, photographs, videos, and text messages on
smartphones. AI is defined as something that can “serve the human by optimizing tasks that
would take too long to be performed by one or a few people” (Santos; Arruda, 2019, p. 728). It
is studied, in this sense, how to program computers to perform tasks that, until then, humans do
better. In the educational field, talking about AI is to create space for discussion, for example,
 31

logical reasoning, programming, algorithms5, computer language, translations, robotics,

artificial neural networks, and more recently, for the first time in history, the writing of
academic and literary texts.
By observing the interface of text message writing on smartphones, we can see an
example of how AI pattern recognition algorithms work, namely, they predict and suggest the
next word to be typed based on what had just been typed. This process is governed by a
statistical language model that distributes probability over the sequences of words. Such
predictive capability exemplifies the core functionality of Predictive AI, the deployment of
Machine Learning (ML) techniques to improve accuracy and efficiency of text entry. The
underlying statistical models, for example n-gram models or more complex neural networks,
analyze vast amounts of text data to learn probable word sequences (Bengio et al., 2003). As
users type, these models apply learned patterns to predict and suggest the next most likely word,
thereby streamlining the communication process. Predictive AI encompasses more than the
improvement of text messaging, also playing a crucial role in applications including email auto-
completion and real-time translation services. These technologies enhance the speed and
intuitiveness of digital communication, making it more efficient and user-friendly.
Predictive AI and Generative AI represent two distinct branches of AI technologies with
divergent functionalities and applications. While Predictive AI increases the efficiency of text
messaging through relevant word suggestions, it differs fundamentally from Generative AI,
particularly in the context of text generation in education. GenAI, exemplified by LLMs like
GPT-4, goes beyond predicting the next word since it generates coherent, contextually
appropriate, and often creative text based on a given prompt. This capability is supported by
Deep Learning (DL) — a subset of ML — whose techniques structure algorithms in layers
forming an artificial neural network capable of assisting in more demanding writing tasks, such
as drafting essays, helping to create lesson plans, or providing personalized feedback on
students’ work. Unlike Predictive AI in text messaging, which relies heavily on existing data to
predict subsequent words, GenAI can synthesize new ideas and articulate nuanced arguments,

5
According to Ferrari and Cechinel (2008, p. 15), in the context of computing, in which machines manipulate data
from an instruction, “an algorithm can be defined as a finite sequence of steps (instructions) to solve a given
problem. Whenever we develop an algorithm, we are establishing a pattern of behavior that should be followed
(a norm of action execution) to achieve the result of a problem. For the development of an efficient algorithm it
is necessary to obey some basic premises at the moment of its construction: (i) define simple and unambiguous
actions; (ii) organize the actions in an orderly way; (iii) establish the actions within a finite sequence of steps [...]
Algorithms are able to perform tasks such as: 1) reading and writing data; 2) evaluating algebraic, relational and
logical expressions; 3) making decisions based on the results of the evaluated expressions; 4) repeating a set of
actions according to a condition”.
 32

making it a valuable tool in educational settings where creativity and critical thinking are
essential (Ahmad et al., 2023; Lancaster, 2023; dos Santos et al., 2023).
To appreciate the full spectrum of AI, it is essential to recognize that AI as a field of
study has a rich history spanning over 70 years. This extensive background encompasses a
broad array of technologies beyond the contemporary focus on predictive and generative
models. Historically, AI research has progressed from early rule-based systems in the mid-20th
century through various paradigms, including symbolic AI, expert systems, and the latest
advances in ML techniques, culminating in today’s sophisticated neural network architectures
(Russell; Norvig, 2021). Each phase has contributed fundamentally to the diverse applications
and robust methodologies observed in current AI technologies.
Nowadays, ChatGPT is considered the most popular LLM based on the algorithm GPT-
4, the unlimited subscription-based version. GPT-4 outperforms its previous models because it
accepts both text and image inputs and emits text outputs. Besides the multimodality abilities,
its capacity to understand and generate more intricate and nuanced content is increased. It can
understand more languages and dialects with greater proficiency. Another extremely important
improvement is the enhancement of safety features, reducing harmful or biased outputs
(OpenAI, 2023a). Its most recent improvements: on September 25, the integration with speech-
to-text and text-to-speech technologies. This makes GPT-4 more accessible and user-friendly
because the user’s voice inputs are converted to text so they can be processed and finally, read
out loud (OpenAI, 2023b), and GPT-4o (“o” for “omni”), the most advanced model to date. It
is a multimodal system, capable of accepting both text and image inputs and outputting text.
This model provides GPT-4-level intelligence and enhances its capabilities across text, voice,
and vision, all while delivering much faster performance (OpenAI, 2024).
The available free version, which now also offers limited access to GPT-4o, is based on
GPT-3.5 and it can be accessed via the official website, or the mobile app. Its predecessor is
the GPT-3, which served as the basis for ChatGPT, has 175 billion parameters. For the purpose
of comprehending this number in perspective, let us factor in account previous models, GPT-2,
considered cutting-edge and surprisingly large when released in 2019, which had 1.5 billion
parameters. Clarifying the understanding about the size of GPT-3, I exemplify with data from
Wikipedia, whose English version surpassed on January 23, 2020, 6 million articles, equaling
only 0.06% of the data trained by GPT-3.
After originally publishing its GPT-3 research in May 2020, OpenAI selected members
of the public and allowed access to the model via an API (Application Programming Interface).
Over the course of subsequent days, samples of text generated by GPT-3 began to circulate
 33

widely on social media. After the release of GPT-3, it became commonplace to come across
reports on the internet with titles alarming about the possibilities and consequences of this kind
of technology for writing. However, extremely far from the serious repercussions that surround
our society nowadays. In 2021, when the article entitled GPT-3: will the world’s most powerful
writing AI replace a journalist? (Zaramela, 2021) was published, it addressed the doubt
regarding the replacement of this professional, described the experience of using this language
model in Portuguese and English, and exemplified with texts initiated by humans and continued
by the AI. Since late November 2022, with the extensive outperformance of GPT-3.5 and GPT-
4 capabilities, many more professions have been impacted by this technology, in particular,
educators and schools. The latest changes that have occurred in society are undoubtedly far-
reaching,
As an English teacher for 27 years, I reflect on the impact of yet another technology in
the history of the language teaching-learning process, particularly the English language, whose
path of methodologies has always been shaped by the development of technologies. Turning to
the context of writing, my professional experience teaches me the affordances of an English
language learner in search of skills, strategies, and the necessary dispositions for writing. Thus,
writing is redefined if we take into consideration aspects such as textual genres, structural issues
of the language, and contents, previously seen as the focus in the learning process of writing
and that by means of language models can be automatically generated. The GPT-3-generated
texts were initiated by humans, but continued by algorithms. With the advancements in NLP,
the GPT-3.5 and most importantly, the ones generated by GPT-4 definitely imposed questions
that were already bothering educators, schools, and educational institutions. Concepts regarding
authorship are part of this debate when it comes to a human-language model writing partnership.
AI then draws attention to ethical and intellectual property implications for writing.
The recent refinement of this technology may further instigate reflections on creativity
in academic writing. Writing academic papers, whether in secondary or higher education, can
be conceived as a source of anxiety and stress for faculty and students (Casanave, 2019; Russell-
Pinson; Harris, 2019). The difficulties lie in choosing the focus to be given to the work as well
as the search for bibliographical references and citations to substantiate the research, to the
writing of the work itself. Students engaged in academic writing need to create texts that are
strictly coherent with what the scientific community expects from them “that will evaluate them
and that they will surely want to join” — a work that implies time and effort, is linked to the
reading of other texts, meets specific standards, and requires the practice of a constant reflective
exercise (Pinto, 2018, p. 9).
 34

Technological changes interfere with the learning of writing, imposing on students not
only the need to learn specific textual genres, but to continuously learn to write across this
variety of technologies, modalities, and contexts (Leu et al., 2016). The authors propose the
development of an expanded perspective on writing through the lens of literacies since
technologies redefine what writing means, as they are developed. In light of these shifts, this
study seeks to explore the intersection of technology and text creation within educational
settings, particularly in the advent of GenAI technologies. The following research questions
guide this inquiry: How does the integration of GenAI technologies, like LLMs, redefine the
concept of writing in educational contexts? What changes in approaches to writing have been
observed following the launch of ChatGPT? What educational and pedagogical issues have
arisen from the integration of AI-text generators in teaching environments? What ethical
concerns are associated with the use of AI-text generators like ChatGPT in education, and how
are these concerns being addressed?
I am guided by the view of literacies of the New London group whose manifesto entitled
A pedagogy of multiliteracies: Designing social futures, published in 1996, recognizes the
importance of graphocentric school literacy, but problematizes it for not being sufficient in the
face of constant changes, local and global, generated by the impacts of information and
communication technologies.
In this pedagogy of multiliteracies, the proposal is the semiotic study of texts, with
different ways of producing, transmitting and consuming them. The group is based on the fact
of the “immense linguistic, cultural and social diversity of today’s world and how this diversity
has become increasingly constitutive of social life” (Kalantzis; Cope; Pinheiro, 2020, p. 117);
and also, on the fact that the written text, as never before possible, integrates with images and
sounds, making possible an increasing multiplicity and integration of ways of constructing
meaning, due to the multiplicity of communication media.
The launch of ChatGPT has undoubtedly marked a turning point for further
transformations in writing. Writing has always been impacted by technologies and this AI-
chatbot highlights the already existing need for discussion taking into consideration ethical
implications, intellectual property, creativity and writing affordances. These are all aspects that
integrate writing as a social practice. To expand my perspective upon this phenomenon, I
understand it is crucial to apprehend the reality that surrounds it toward the possibility to
contribute to the generation of new knowledge both scientifically valid and socially relevant
(Petraglia; Moraes, 2022). The reality perceived and investigated in this study is the concept of
 35

writing in the GenAI context, undergoing one more of its various transformations along human
history.
Considering this context, the object of this study is the concept of the phenomenon of
writing impacted by GenAI in the field of education. My thesis aims to support the proposition
that the use of GenAI broadens the concept of the phenomenon of writing as a social technology.
The purposes of my research are to develop a concept of writing in the context of LLMs and
GenAI; to reflect upon the impact of this technology on education; and to reflect upon how to
write and how to teach writing. The general objective is to apprehend the concept of the
phenomenon of writing impacted by the adoption of GenAI in educational contexts.
Specifically, this study aims to examine the approach to writing conveyed on internet-based
platforms, particularly after the launch of ChatGPT; and to explore educational, pedagogical,
and ethical issues related to the impact of AI-text generators, especially ChatGPT, raised on
internet-based platforms. The research utilizes as data sources documents such as digital
newspapers news, social media posts, YouTube videos, websites, and podcasts (see Figure 10).
 36

Figure 10 – Research concept map

Source: created by the author (2023).

 37

Bearing the above in mind, I present Complex Thinking as an epistemological and

methodological framework for carrying out this investigation. Through the lens of Complex
Thinking, the object of my study is considered a phenomenon. The term phenomenon is defined
as life experiences that emerge from our social and cultural interactions (Ravagnoli, 2017).
Complex originates from the Latin term ‘complexus’ and signifies ‘woven together’.
Considering Morin (2015) and adopting a thought “the least mutilating possible”, respecting
the investigation demands, proper to scientific knowledge, and the reflection requirements
proposed to philosophical knowledge.
Morin (2000) understands that the meaning of words changes according to the context
used, and in linguistics, the meaning of a text is understood only by the context in which it is
found. Thus, words and information, writing itself, might not be understood if isolated from the
context of our society, our knowledge, and our culture. We should therefore be motivated by a
principle of thought that allows us to connect things that seem separate to us in relation to each
other. The study of writing should not be regarded as an overly specialized field in isolation.
Based on Complex Thinking, the study of writing cannot be done in a way that does not consider
the aspects inherent to it in the school environment but also cannot happen in a disconnected
way from the social and communication scenario. Education is integrated in the context of
social and technological changes whose uses impact on the evolution of writing. According to
the holographic principle, “society as a whole appears in each individual, through language,
culture, and norms” (Morin, 2000, p. 16). In the recursive ring principle, products and effects
are both producers and causers of what produces them: “Human individuals produce society in,
and through, their interactions, but society, as an emergent whole, produces the humanity of
these individuals by taking language and culture apart from them” (Morin, 2000, p. 16). Such
statements explain the transition of the various concepts of what is meant by writing, from
writing on paper to the computer screen; after the internet, appropriating multimodal
characteristics and, later, with social networks, as a social practice, justifying the relevance of
researching the meanings about writing taking into account the scenarios that technologies
integrate socially.
Thus, I justify the insertion of my proposal in the research line Technologies,
Languages, and Education of Programa de Pós-Graduação em Educação (PPGED), whose
studies focus on problematizing and investigating the relationships between education and
communication processes, considering oral, written, analog and digital languages in educational
practices, in face-to-face and distance learning modalities. My motivation for pursuing this
research theme is due to both my personal and professional interests in English language
 38

teaching and learning and also, because of my academic background. My contact with the
English language as a learner started when I was 10 years old in a language course, in 1986,
and as a teacher, when I started teaching in language schools in 1996. In 2003, even though I
held a degree in Dentistry, completed in 2001, at UFS, and, at the time, I had already been
working as an English teacher for 7 years, I decided to start a degree in English at UFS. In the
period, between 2011 and 2020, when I worked as a pedagogical coordinator for Elementary
and Middle Years, in a bilingual school (Canadian education system) in Aracaju, I expanded
my contact with the English language by guiding the implementation of English programs:
Language Arts, grades 1 to 9, and also Science and Math, since they are subjects taught in
English in this institution. I realized my growing connection with the field of Education, which
led me to hold a Post-Baccalaureate Diploma in Education at the University of Winnipeg, in
2014. Upon my return to UFS, as a substitute professor at the Department of Modern Languages
and coordinator of subjects from the Modern Languages Department of the Distance Higher
Education Center, Centro de Educação Superior a Distância (CESAD), between 2017 and
2019, I taught subjects such as Written Expression in English I; English Language II, III, IV
and VIII, among others, and I worked as a tutor for English Language undergraduate students
at CESAD, advising them in the writing of their scientific articles as the final paper. Allied to
the challenges and experiences of my teaching practice, I also see as reasons for my interest in
the theme: the writing of my Master’s thesis in Education, whose title is Representations of
English teachers on technologies and their relationship with teaching practice, defended in
2020, in addition to my participation as a member of the study and research group Technologies,
Education and Applied Linguistics (TECLA), scenarios that constantly reinforce my interest in
the Education and Communication research lines; Languages and Technologies in Education;
Language and Society.
In the context of Complex Thinking, the relationship between subject and object is by
interdependence. This implies that the subject’s perception and understanding of the
phenomenon investigated do not seem disconnected. With reference to this epistemology and
methodology, the researcher is treated as a research participant. As a result, it is relevant to
describe my path since I started investigating this object as well as the in-depth look at it while
close investigation and interaction during collaborative writing. All that allowed me to view my
object of study from different perspectives (see Figure 11).
 39

Figure 11 – Scientific publications and presentations

Source: created by the author (2023).

 40

The credibility of this research was established from the onset of my doctoral journey,
through continuous guidance from my advisors, reflections stemming from discussions in our
research group meetings, and oral communications of the ongoing project at conferences,
seminars, and panels, as outlined in Appendixes I and II.
As theoretical bases to reference this study are Morin (2015), Petraglia (2020) and
Petraglia and Moraes (2022) for Complex Thinking. Bazerman (2023), Boa Sorte (2018),
Fischer (2021), Dibbell (2010), Christensen (2012), Kist (2005), Leu et al. (2016), Kalantzis
and Cope (2024) and MacArthur et al. (2016) as authors for writing. In the area of GenAI and
writing, Boa Sorte et al. (2021), Kalantzis and Cope (2024), dos Santos, Boa Sorte & de Oliveira
(Forthcoming). The theorists for GenAI and ChatGPT are Santaella (2023) and Wolfram
(2023). In the sphere of multiliteracies, the authors are Kalantzis, Cope and Pinheiro (2020),
Cope, Kalantzis and Searsmith (2020), Cope and Kalantzis (2000) and Kalantzis and Cope
(2023, 2024).
Methodologically, this study is described as qualitative because it emphasizes the
interconnected and emergent properties of human experiences. This qualitative research is
conducted in light of Complex Thinking (Morin, 2015) as epistemology and methodology in
order to pursue this path of uncertainty, innovation, provisionality, and randomness. This is a
documentary and bibliographic research as regards data sources and gathering procedures. The
data are documents selected on internet-based platforms, whose narratives address GenAI and
its relation with education and writing as well as ChatGPT and writing. In view of this, I present
Analytical Entwining integrated with GenAI (adapted from Boa Sorte et al., 2023; dos Santos;
Boa Sorte; de Oliveira, forthcoming) as the method of interpretation for this investigation.
While upholding ethics and scientific rigor, its flexible nature provides researchers with four
to-and-fro movements as they engage with data.
I continue this study presenting my methodological path in the following section. I
describe the characteristics of Complex Thinking (Morin, 2015), the epistemology and
methodology I follow, as well as the data gathering procedures and the method of analysis. In
section two, I narrate the change of course, considered inevitable from the Complex Thinking
perspective and given the current context of the object of my investigation, and also propose a
method of analysis integrated with GenAI: Analytical Entwining integrated with GenAI
(adapted from Boa Sorte et al., 2023; dos Santos; Boa Sorte; de Oliveira, forthcoming). In
section three, I present and discuss the theoretical constructs that helped me achieve the first
specific objective: to examine the approach to writing conveyed on internet-based platforms,
particularly after the launch of ChatGPT. In section four, I present and discuss the theoretical
 41

constructs that helped me achieve the second specific objective: to explore educational,
pedagogical, and ethical issues related to the impact of AI-text generators, especially ChatGPT,
raised on internet-based platforms. Finally, in my final remarks section I present the concept of
the phenomenon of writing impacted by GenAI that emerged from this research conducted
through the lens of Complex Thinking principles (Morin, 2015). Referring to section one
epigraph, I pursue this path having in mind aspects Morin (2015) acknowledges as essential for
doing research: uncertainty is inherent to human existence and understanding; simple
explanations or models cannot apprehend realities; and the truth is not found in total
understanding, but perhaps in the recognition of the limitations of our knowledge.
 42

SECTION 2

METHODOLOGICAL PATH6

6
Image generated using OpenAI’s DALL-E image generation tool on November 11, 2023. The prompt used for
generation was: “Generate an image that represents a researcher in her initial path of study. Her investigation
focuses on education, writing, artificial intelligence, and ChatGPT.”
 43

2 METHODOLOGICAL PATH

“Do not forget that reality changes,

do not forget that the new may happen,
and it will happen no matter what”
(Morin, 2015, p. 83).

In navigating research paths, it is recommended to maintain an open mind towards

necessary, gradual, but oftentimes, abrupt transitions that scientific research may undergo in
the course of time. Whenever actions are carried out, methodological choices have to be made
(Morin, 2015). Choosing initial and subsequent strategies in scientific investigations implies
awareness of the existence of risks and uncertainties as constant challenges.
To fully understand the implications of these choices, we need to revisit the concepts of
epistemology and method (Petraglia, 2022): Epistemology, from the Greek: episteme
(knowledge) and logos (study), is the philosophical domain dedicated to how science and
humans acquire and justify their knowledge. It embodies the philosophy of science or the theory
of knowledge, aiming to elucidate its origins, logical foundations, structure, and value. An
epistemological perspective seeks to identify the essential requirements and conditions for
understanding a statement. Method, from the Greek term: methodos (path or way) refers to
means or manner, a set of organized strategies and/or procedures. It includes the methodology
with its techniques and materials that we employ to reach a point, achieve a goal, or address a
problem.
As the epistemological and methodological framework of my study, I follow Complex
Thinking proposed by Morin (1990, 2015). In this section, I address indispensable elements
that support me throughout the designing of this investigation and the building of my path in
light of this epistemology. First, I present the fundamental principles, also known as cognitive
operators, which continuously shape my journey as a researcher guided by Complex Thinking.
In the second subsection, I describe the nature of research and the questions that help us navigate
the methodological procedures along the study. In the two following subsections, I narrate the
course change, expected due to the uncertainty trait within Complex Thinking, and the
designing of the new path in my research journey. I conclude this section by presenting the
method of interpretation for this investigation: Analytical Entwining integrated with GenAI
(adapted from Boa Sorte et al., 2023; dos Santos; Boa Sorte; de Oliveira, forthcoming), and
explaining the four to-and-fro movements I follow as I collaborate with ChatGPT to engage
with my research data.
 44

2.1 Complex Thinking and its core principles

Upon first examination, the term “complex” tends to lead to erroneous understanding.
The notion that ‘complexity’ primarily expresses our discomfort and confusion suggests that
the word carries a heavy semantic burden and encapsulates uncertainty and disorder. Complex
Thinking is an approach to understanding the world that acknowledges, embraces, and attempts
to comprehend the intricate web of relationships, interactions, and interdependencies that
constitute reality (Morin, 2015).
Complex Thinking stands in contrast to reductionist thinking, which simplifies and
separates phenomena into discrete, isolated parts, often neglecting the richness of connections
and the unpredictable dynamics that emerge from the whole. It rejects the oversimplification of
reality, highlighting that complexity cannot be reduced to a single law or simple idea. Instead,
he argues for acknowledging the multidimensional, intertwined nature of reality, which cannot
be fully apprehended by reductive thinking (Morin, 2015).
Although Complex Thinking does not offer a prescriptive solution for unexpected
events, this epistemology encourages alertness, responsibility, and intellectual rigor, reminding
us that changes are inevitable and should be anticipated. In accordance with what Morin (2015)
referred to in the epigraph of this section, the presence of elements such as randomness,
indetermination, and uncertainty is the absolute certainty in life. “In some ways, complexity
always relates to randomness” (Morin, 2015, p. 35). So, we need strategies in order to pursue
our research journey. It is not uncommon to mislead ourselves by running on autopilot and
performing planned actions to solve simplified issues with simplifying thinking. This is
essential in order to carry out research: steps planned to reach goals as long as random situations
do not arise during the course of the study. The need for strategies aligned with Complex
Thinking represents decisions and challenging choices leading to risky and unclear actions.
However, Complexity focuses on an initial point in an investigation, allowing for broader
actions that are less restricting and limiting, and simplifying perspectives.
This approach encompasses a range of principles, acknowledging the coexistence of
order and disorder as inherent to human essence and experience. Importantly, Complex
Thinking does not oppose clarity, order, or determinism. Nevertheless, what distinguishes this
epistemological and methodological foundation from reductive paradigms is the fact that these
three elements, while valuable, are seen as insufficient, since planning does not assure the
attainment of discovery, knowledge, or action.
 45

I present the fundamental principles that continuously shape my journey as a researcher

guided by Complex Thinking. The first, the dialogic principle, suggests that seemingly
opposing elements or concepts can exist in a mutually beneficial, complementary relationship
(Morin, 2015). This principle is a cornerstone in the study of complex systems, where dualities
are not only inevitable but also essential for understanding the true nature of a phenomenon.
The dialogic principle encourages a form of complex thinking that acknowledges and embraces
these dualities, rejecting oversimplified views of reality that categorize things as merely “black
or white”. Instead, it promotes a nuanced understanding where contradictions can coexist and
inform each other.
The recursive principle plays a crucial role in conducting research that adopts Complex
Thinking (Morin, 2015). This principle is central to this epistemological and methodological
approach because it challenges linear and reductionist ways of understanding a phenomenon.
The recursive principle refers to a process where the products and effects are simultaneously
cause and effect; product and producer. In simpler terms, it describes a situation where certain
processes or phenomena are both the cause and the effect of each other, creating a continuous
loop of interaction and influence. In sociology or anthropology, recursive processes are
observed in how cultural norms shape individual behavior, and in turn, how individual
behaviors can transform cultural norms over time.
Holographic is the third principle (Morin, 2015). This principle highlights the
interconnectedness and interdependence of parts within a whole. Changes in one part can have
significant implications for the whole system. It suggests that in complex systems, every
component is not merely a part but also contains the entire system. This concept is inspired by
the nature of holograms, in which every piece of the hologram holds a complete image, though
in a reduced form. Unlike traditional perspectives that view parts and wholes as distinctly
separate, the holographic principle suggests that each part reflects the whole. This means
understanding a part of the system requires considering its relation to the whole system. In
addressing complex issues, this principle suggests that solutions should not only focus on parts
of the problem but also consider the systemic nature of the issue, understanding how each part
relates to and impacts the whole. In education, this principle emphasizes the importance of
understanding individuals or units in the context of the larger system, recognizing that each
element reflects and contributes to the larger picture. Morin’s holographic principle thus
provides a crucial lens for understanding complexity in natural, social, and cognitive systems,
advocating for a more interconnected, systemic view of the world.
 46

In Complex Thinking, each principle contributes to the understanding of the nature of

phenomena experienced in life. The holographic principle suggests that each part contains the
whole, while the recursive principle emphasizes the self-producing and self-organizing nature
of systems. Additionally, the dialogic principle recognizes the coexistence and interaction of
complementary and contradictory forces. Collectively, these principles depict processes or
phenomena as complex, adaptive, and evolving entities, where understanding emerges from
recognizing the interconnections among the parts, rather than from isolated interpretation,
forming the notion of system.
Complex Thinking, therefore, is not solely a cognitive process (see Figure 12) but an
intellectual stance that challenges the thinker to embrace the multifaceted, intertwined, and
often contradictory nature of reality. It is a call to move beyond the comfort of certainty and
simplicity and engage with the full spectrum of complexity that defines our world. Morin (2015)
advocates for a kind of thought that is agile, adaptable, and capable of navigating the inherent
complexities of the world we inhabit, with an aim not to eliminate complexity but to better
understand and work within it.
 47

Figure 12 – My research in light of Complex Thinking7

Source: created by the author using DALL-E (2023).

What constitutes the reality of this study is the complex tangled web of relationships,
interactions, and interdependencies among writing, schools and education, AI-text generators,
and societal contexts represented by the phenomena of digital newspaper news, social media
posts, YouTube videos, podcasts, and websites. In consideration of the unpredictability and
transdisciplinary of this reality, I aim to apprehend the concept of the phenomenon of writing
impacted by the adoption of GenAI in societal contexts in the field of education in light of the
Complex Thinking to achieve the following purposes: to develop a concept of the phenomenon
of writing in the context of LLMs and GenAI; to reflect upon the impact of these technologies
on society and education; and to reflect upon how to write and how to teach writing.

7
Image generated using OpenAI’s DALL-E image generation tool on November 11, 2023. The prompt used for
generation was: “Generate more images taking this *previous prompt into account but also bearing in mind that
epistemology of Complex Thinking does not separate contexts in layers since it believes it is all connected like
systems that integrate with one another. This means it is hard to know when one context or system starts or ends”.
*Previous prompt: I’m a researcher in the area of education, languages and technologies and I’m writing a
scientific paper. I want to include an image in this paper that, in light of Complex Thinking (Edgar Morin), shows
I’m organizing a section of this paper starting from a macro contextualization of social, political and economic
contexts. Then, I focus my view on societal contexts in the field of education. So, I address AI algorithms and
Generative AI impacting elections, free press and democracy for my social, political, and economic contexts. In
addition, I focus on the impact of generative AI, Large Language Models on writing regarding assessment and
teaching.”
 48

2.2 The nature of research

In view of Complex Thinking, researching means to construct or revise, and synthesize

knowledge on a specific subject, issue, process, or observed phenomenon, concerning a
particular problem that grabs our attention and compels us to want to delve deeper into its
investigation. This accumulated body of knowledge becomes a scientific narrative about the
researched phenomenon, thus bringing order to the previous and inner chaos that aroused our
curiosity and prompted us to investigate. However, the knowledge developed here goes beyond
the everyday understanding of the objectified reality. It belongs to what we term as scientific
knowledge because it stems from scientific research carried out in a more rigorous, consistent,
organized, and coherent manner.
In other words, it arises from inquiry, a more systematized, empirical, validated, and
refined investigation, regardless of its quantitative, qualitative, or mixed nature. What’s crucial
to note here is that producing scientific research always requires a certain discipline and a
particular scientific rigor, ensuring that events and facts are not left to fate. And today, we
understand that the more a country invests in scientific research, the more it progresses socially,
economically, technologically, and culturally speaking (Petraglia; Moraes, 2022). As always,
it’s essential to ensure that the translated content retains your intended meaning and fits within
the context of your article.
Complexity does not conform to predefined frameworks or wholly standardized
practices. This does not mean that we are devoid of tools, procedures, or techniques. Rather, it
suggests that we should not prioritize method over experience, as there is no single route or
uniform way to think, investigate, invent, or gain knowledge. However, in order to conduct
research in light of Complex Thinking, some questions have to be answered (see Figure 13)
(Petraglia; Moraes, 2022).
 49

Figure 13 – Complex Thinking: research guiding questions

Source: created by the author based on Petraglia and Moraes (2022).

Embracing Complex Thinking implies that the nature of research should not be strictly
framed and linear, as it is based on dialogic, recursive and holographic principles, and the notion
of system. However, the research guiding questions are meant to be viewed as helpful to the
methodological procedures, enabling researchers to navigate through an investigation .

2.3 The change of course

In light of Complex Thinking, the journey of a research is seen as one that unfolds while
one is walking. Therefore, where at any given moment, in response to emerging situations, there
might be a need for a new decision: to embark on a new path or to take a different fork in the
road (Morin; Motta; Ciurana, 2003).
When I started my doctoral program, LLMs were already accessible to the Brazilian
public. AI-writing assistants based on GPT-3 algorithms could be utilized temporarily at no
cost to the user. Although there were serious worries regarding ethical aspects (see Figure 14)
surrounding the adoption of AI platforms for text generation, they seemed to be preliminary
concerns considering what the media reported. Prior to the launch of ChatGPT, the topic
appeared to be in the initial phase, given the low frequency and quantity of media publications.
 50

Figure 14 – GPT-3 in journalism

Source: Zaramela (2021).

The article entitled “GPT-3: Will the world’s most powerful writing AI replace a
journalist?” was published approximately six months after the launch of GPT-3. Within this
publication, the model is portrayed as a groundbreaking advancement in the field of GenAI,
with significant implications for journalism and other professional fields. It highlighted the
prevalent fear that this technology might signal the end of the journalism profession. It took an
additional eight months for the Brazilian television media to discuss the topic on a national
scale (see Figure 15).

Figure 15 – AI: Fantástico News report

Source: Globo (2021).

 51

The news report is about a Canadian who participated in a project during the COVID
pandemic. The project, called December, used AI technologies to simulate exchanging
messages with his fiancée, who had passed away 8 years before this report. This news report
exposes the risks pointed out by specialists in the development of increasingly advanced AI and
questions the ethical limits of its use. The concerns seem to revolve around the ethical
implications and potential psychological effects of using AI to simulate interactions with
deceased individuals.
In 2021, I had defined my study goals and I was working on my research project, reading
the literature, designing the field research, and familiarizing with UFS’s Ethical Committee in
Research, Comitê de Ética em Pesquisa (CEP), updated ethical procedures. As a researcher, I
value the ethical aspects when conducting scientific research. Due to that, in that same year, I
submitted the research project to the Institutional Review Board of the Federal University of
Sergipe in June and only after its approval8, data collection was meant to begin. The approval
was given at the end of year 2021, enabling me to commence field research.
These plans were placed on hold after I was granted a PDSE CAPES scholarship. In
September 2022, I relocated to Richmond, Virginia (USA), to begin my “Sandwich” doctoral
program at Virginia Commonwealth University (VCU) under the supervision of Prof. Dr.
Luciana C. de Oliveira. In consultation with my advisor, we decided to delay the field research
so I could experience this unique educational opportunity to the fullest and expand my
understanding of the object of my study.
During my interactions with faculty and students at the School of Education (VCU), I
frequently shared information about my research. Initially, responses were a combination of
surprise, curiosity, and confusion, as the theme was unfamiliar to those I engaged with. This
perception gradually shifted subsequent to the introduction of ChatGPT on November 30, 2022.
By December 2022, I faced a crucial decision regarding the need for financial support
to utilize a GPT-3 platform. This platform was essential for the continuation of my research
post the “Sandwich” program and the initiation of data collecting process. During data
collection, the participants of an academic writing course would use a platform or application
with the GPT-3 language model, such as ShortlyAI9. The selection of the natural language
model, the GPT-3 writing assistant, was pivotal due to its potential impact on the research

8
This research was approved by the Institutional Review Board of the Federal University of Sergipe-UFS (CAAE
50114121.2.0000.5546) and received substantiated opinion n. 5.173.020.
9
ShortlyAI is a text generator application, in which the user just types in initial sentences and by clicking the
“write for me” button, the AI writer takes over the writing. Available at: [Link] Accessed
on: May 12 2021.
 52

outcomes. This thinking process involved rigorous inclusion and exclusion criteria to select the
most suitable AI writing assistant from the various options available in the market. The
selection of the platform posed a particular challenge due to the vast number of options available
and the changing criteria throughout the study.
In the context of selecting an AI text-generator, mind mapping proved instrumental. A
mind map visually organizes the selection criteria into color-coded branches (Buzan, 2024).
The green color corresponds to the criterion of the content model, indicating the GenAI’s ability
to handle various writing styles and formats. Following the clockwise direction, the pink branch
represents the types of accounts available—individual or group. Purple is used to denote the
version of the software, distinguishing between free and paid options, which is critical for
budget considerations. The gray highlights the company producing the technology, essential for
understanding the support and innovation behind the product. The need to select the most
advanced model is highlighted in the red branch. Finally, the blue branch represents the date of
launch, indicating the software’s currency and relevance in the fast-evolving field of AI (see
Figure 16).

Figure 16 – AI text generator selection criteria

Source: created by the author (2023).

This approach serves to clarify the decision-making process and also highlights the
strategic thinking involved. At the end of 2022, after progressing through many relevant steps
in developing my research, my advisor and I began to recognize the need to reconsider my
research trajectory.
This realization was further catalyzed by the substantial impact of ChatGPT, which
achieved 100 million users within just two months of its release. Additionally, the release of
 53

GPT-4 by OpenAI on March 14, 2023, and the announcement of the potential launch of GPT-
5 during the same event, compelled us to reassess the research direction. These developments
highlighted the rapid advancement of GenAI technologies and their implications for my study.

2.4 Redesigning my path

A journey of understanding certainly begins with an intent and anticipates an end. It is

crucial to recognize the role of theory and its relationship to method. Theory is not knowledge
itself but rather a facilitator of knowledge, which corroborates Morin’s movement. It is not an
endpoint but an opportunity for commencement, which is a recursive movement. Theory is not
an answer but provides a means to address a problem. A theory only serves its cognitive purpose
and comes to life through the active mental engagement of the individual. This active
engagement, fostered through the dialogic and recursive interactions (Morin, 2015), is what
makes the concept of method essential (Morin; Motta; Ciurana, 2003, p. 23-24).
The flexible nature of Complex Thinking expresses a convergence of events stemming
from the intrinsic complex features inherent to the phenomena in my study: writing within the
context of GenAI technologies. These technologies, capable of generating human-like texts,
impact a variety of societal contexts in the field of education. In response to the rapid
technological advancements in this area, I have redesigned my research trajectory. This
redesign, driven by the Complex Thinking research guiding questions, directly informed my
new methodological path, illustrating the holographic approach, where each part of my research
reflects and contains the complexities of the whole (see Figure 17).
 54

Figure 17 – Complex Thinking: my research guiding questions

Source: created by the author based on Petraglia and Moraes (2022).

Following these questions guided the redesign of my method. The process required time,
reflection, and alertness to new societal aspects influenced by the massive impact of GenAI
technologies in the field of education shown on internet-based platforms. Consequently, I found
it necessary to continue working with the Complex Thinking guiding questions to attain a
clearer understanding of my new path (see Figure 18).
 55

Figure 18 – My guiding questions

Source: created by the author based on Petraglia & Moraes (2022).

 56

Within the complex perspective, theory acts like a cognitive blueprint that leaves lasting
impressions, while the method requires strategy, initiative, invention, and artistry for effective
implementation (Morin; Motta; Ciurana, 2003). There is a recursive relationship between
method and theory: the method is derived from theory and, in turn, revitalizes the theory itself.
A theory that possesses complexity can maintain its intricate nature only through constant
intellectual recreation. Without continual innovation, such theories are at risk of degeneration
into oversimplification. Left unattended, theories tend to become flattened, losing their
multidimensionality and becoming static. From a complexity standpoint, theory is
indistinguishable from method; both are essential components of complex knowledge.
In light of Complex Thinking, and following a deep engagement with the guiding
questions of my research, the new design of my methodological path became more explicit (see
Figure 19). This mind map summarizes the study in terms of four methodological aspects: type
of research, research participant, data gathering, and data analysis (Gil, 2002).

Figure 19 – Designing my path

Source: created by the author (2023).

This visual representation shows the change of course in data source towards a
documentary study motivated by the increasing presence of GenAI in the media, its rapid
advancement, and its impacts on writing and education. The documents selected for this
qualitative research are primarily sourced from internet-based platforms, with collection
commencing at the start of my doctoral program in September 2020. The first relevant
 57

document addressing the theme was from the British digital newspaper, The Guardian. This
article, mentioned in the introduction (see Figure 6), appeared approximately three months after
OpenAI released the algorithm GPT-3 on June 11, 2020. Although my doctoral program had
not officially started at that time, I began examining the impacts of earlier versions like GPT-
2. What naturally transpired over these years was the accumulation of media that addressed the
object of my study. Using the note-taking tool Obsidian, I saved links to YouTube videos,
podcasts, websites, and digital newspaper articles. Additionally, I archived social media posts
using Instagram’s saved feature.
I refer to these documents as ‘screenshots of the reality I see’. They have become central
to interpreting the interconnections among writing, education, and GenAI technologies, all of
which are phenomena in their own right. Phenomena are seen as interactions representing
experiences and perceptions of reality (Ravagnoli, 2017). Therefore, these documents can also
be viewed as phenomena helping to apprehend the concept of writing by the adoption of GenAI
in societal contexts in the field of education. Having detailed the methodological path, the
subsequent subsection outlines the analytical method employed on the collected data.

2.5 Method of analysis

In this subsection, I present the method of analysis for this study, Analytical Entwining
integrated with GenAI, adapted from Boa Sorte et al. (2023) and dos Santos, Boa Sorte & de
Oliveira (forthcoming). First, I explain Analytical Entwining (Boa Sorte et al., 2023), as it
forms the foundation for this analytical method. Next, I detail the data analysis strategies
developed and describe how I leveraged ChatGPT to engage with the research data.
Analytical Entwining (Boa Sorte et al., 2023) has emerged from research carried out by
members of the research group mentored by my advisor, Prof. Dr. Paulo Boa Sorte,
Technologies, Education, and Applied Linguistics (TECLA), from the Federal University of
Sergipe. This analytical perspective results from the interlacing between the theoretical-
methodological contributions of Teacher Research (Freeman, 1998) and Coding Process
(Auerbach; Silverstein, 2003). It addresses the need to expand the possibilities for instruments
and techniques for data gathering in Education and Applied Linguistics research, particularly
considering the potential of digital technologies, and more recently, AI.
This method consists of four steps named analytical movements (see Figure 20). These
movements serve as guiding stages that help researchers manage large datasets while
consistently upholding ethical standards and scientific rigor. The method’s flexible nature
 58

allows for to-and-fro movements, enabling researchers to achieve study goals and respond to
research questions as they engage with data. Simultaneously, Analytical Entwining permits
researchers to reconsider and reshape their research trajectory in response to new insights and
reflections. During the process, carrying out analysis and redesigning the path tend to generate
anguish and uncertainty originating from data. Ultimately, this method positions the researcher
at the core of the study, inherently interconnected with the object of investigation.
Given these considerations, it is discernible how selecting Analytical Entwining (Boa
Sorte et al., 2023) naturally integrates into the epistemological and methodological approach
selected for this study. It properly aligns with Complex Thinking (Morin, 2015) through its
engagement with the subject-object relationship, a key epistemological principle in research
due to its methodological implications. Furthermore, this method of interpretation does not aim
to obtain results perceived as singular and decontextualized truths whose conclusions seek to
define a research object exhaustively. Conversely, it realizes the subject as “historically situated
within a specific context, and their perspective emerging from social constructions” (Boa Sorte
et al., 2023, p. 26). It also agrees with the notion that knowledge is changeable and certainties
are not permanent according to Complex Thinking (Petraglia; Moraes, 2022). Consequently,
these are the reasons why I integrate Analytical Entwining into the conduct of my investigation.
To describe this analytical approach, I refer to its visual representation (see Figure 20).
This image depicts, in a clockwise direction, the four to-and-fro movements intended to guide
researchers while interpreting data: locating relevant excerpts, recognizing repeated ideas,
categorizing revealed themes, and elaborating theoretical constructs (Boa Sorte et al., 2023).
The authors note that the choice of the term movements instead of steps is intended to highlight
the absence of linearity when an analysis is being carried out, as shown in the illustration.
 59

Figure 20 – Analytical Entwining

Source: Boa Sorte et al. (2023), my translation.

The analysis begins with the researcher “locating relevant excerpts” in each document,
interview transcript, or any other data-gathering instruments. For the purposes of this research,
I focus on documents, as they are the primary source of data for this study. Documents are
stored in folders on a computer, although working with printed copies is also an option. During
this initial movement, the investigator highlights passages that contain ideas significant to the
research objectives or questions. This analytical movement involves decision-making, guided
by questions such as: Does this excerpt relate to my research objectives? Does it help me
understand the object of my study better? Does it clarify my research questions? Is this excerpt
relevant even if I cannot identify the reason? This last question is important to remind
researchers to either listen to the participants or be attentive to other meaningful aspects
regarding the object of study.
Following this, the researcher opens the file containing relevant excerpts from
Document 1 and compiles a new file termed the ‘list of repeated ideas’ for that document. The
researcher examines these selected relevant excerpts in detail, identifying passages similar to
the initial relevant excerpt aiming at “recognizing repeated ideas”. These findings are then
highlighted, copied, and consolidated into the repeated ideas file. This process is replicated for
each document, forming the initial building blocks for developing a theoretical narrative.
 60

The list of repeated ideas from all documents is organized into a single file, known as
the ‘macro list of ideas’. This list serves as the foundation for the thematic organization of data.
The movement, “categorizing revealed themes”, involves grouping recurring ideas into
coherent categories. This third stage starts with an initial theme, with each theme representing
a cluster of recurring ideas. The researcher identifies conceptual similarities among the repeated
ideas, forming the basis of each theme. This basis is derived from both literature review and
data analysis, as the researcher gains insight into each document. When naming themes, the
researchers are advised to avoid jargon and prefer titles that clearly express the union of the
repeated ideas for easy understanding.
Throughout the four to-and-fro analytical movements, the researcher continuously
reflects on how each element relates to the research questions and objectives, while also
considering significant outlier information from the documents. The researcher’s cognitive
style determines the approach mostly described as a subjective and interpretive process since it
does not impose strict rules for identifying relevance or grouping ideas. These are essential
reminders for the conclusion of the analysis when the themes are grouped with the objective of
“elaborating theoretical constructs”, an example of a Complex movement.
To exemplify the application of Analytical Entwining, I analyze documents, including
digital newspaper articles published after November 30, 2022, the release date of ChatGPT.
The selection criteria for these documents include alignment with research goals and
publications within the first two months post-release, a period chosen due to this AI chatbot’s
unprecedented global impact, evidenced by reaching a user base of 100 million in record time
(see Figure 21).
 61

Figure 21 – ChatGPT: 100 million active users

Source: DemandSage (2023).

Considering these inclusion criteria, I begin the analysis by examining eight digital
newspaper articles (see Table 1), identifiable through color-coded excerpts (see Tables 2 and
3). Prior to initiating the analytical movements, it is essential to have in mind the general and
specific objectives to be achieved in the investigation. The probability of becoming entangled
in the abundance of data is a concern that Freeman (1988) and Auerbach and Silverstein (2003)
point out. Therefore, the general objective of my research is to apprehend the concept of the
phenomenon of writing impacted by the adoption of GenAI in societal contexts in the field of
education. Additionally, the specific objectives are to examine the approach to writing
conveyed on internet-based platforms, particularly after the launch of ChatGPT; and to explore
educational, pedagogical, and ethical issues related to the impact of AI-text generators,
especially ChatGPT, raised on internet-based platforms.
 62

Table 1 – Data: digital newspaper articles

Article Title/Author Publication Date Newspaper name

What is the AI chatbot phenomenon ChatGPT and could Dec. 5, 2022 The Guardian
it replace humans? (Samantha Lock)
TechScape: meet ChatGPT, the viral AI tool that may be Dec. 6, 2022 The Guardian
a vision of our weird tech future (Alex Hern)
The college essay is dead (Stephen Marche) Dec. 6, 2022 The Atlantic

ChatGPT: new AI chatbot has everyone talking to it Dec. 7, 2022 BBC

(Chris Vallance)
The end of high-school English (Daniel Herman) Dec. 9, 2022 The Atlantic

AI-assisted plagiarism? ChatGPT bot says it has an Dec. 31, 2022 The Guardian
answer for that (Alex Hern)
Don’t ban ChatGPT in schools. Teach with it. (Kevin Jan. 12, 2023 The New York
Roose) Times
ChatGPT: student builds app to sniff out AI-written Jan. 13, 2023 BBC News
essays (Nadine Yousif)
Source: created by the author (2023).

I start the analysis by the initial contact with the data when “locating relevant excerpts”,
considering the objectives of this research. I number each of my documents and begin the
second analytical movement by choosing the idea number one and continue “recognizing
repeated ideas” both in document one and in the other seven documents. The idea number one,
“It is an AI chatbot phenomenon […]” guides the identification of other terms that express the
content of the selected initial idea (see Table 2).

Table 2 – Repeated ideas: initial theme

“It is an AI chatbot phenomenon […] The tool has impressed experts […]”
[…] AI chatbot capable of understanding natural human language and generating impressively
detailed human-like written text.”
“OpenAI’s new chatbot isn’t a novelty. It’s already powerful and useful – and could radically
change the way we write online […]”
“OpenAI released an advanced chatbot named ChatGPT that has spawned a new wave of [...]
It is frankly amazing.”
“The results have impressed many who’ve tried out the chatbot.”
“ChatGPT is a freakishly capable tool that landed in their midst with no warning, and it performs
reasonably well across a wide variety of tasks and academic subjects.”
“This technology is only going to get better and better, AI is here to stay. This is the future,”
The popularity of ChatGPT since its launch has been met with alarm, including from some US
schools who have blocked it on their servers in a bid to prevent students from cheating.”
“The arrival of OpenAI’s ChatGPT, a program that generates sophisticated text in response to
any prompt you can imagine, may signal the end of writing assignments altogether—and maybe
even the end of writing as a gatekeeper, a metric for intelligence, a teachable skill.”
Theme 1: Writing is impacted by GenAI technologies, especially ChatGPT.
Source: created by the author (2023).
 63

The repeated ideas in Table 2, expressed by words highlighted in bold and represented
by the excerpts: “[...] AI chatbot phenomenon”; “[...] It’s already powerful and useful – and
could radically change the way we write online […]”; “ChatGPT is a freakishly capable tool
that landed in their midst with no warning […]”; “[…]AI is here to stay. This is the future […]”
address ChatGPT’s capabilities to generate texts and its reactions in societal contexts. By
recognizing these repeated ideas, I categorized the first theme, entitled writing is impacted by
GenAI technologies, especially ChatGPT. The analytical movements of identifying a second
repeated idea continues (see Table 3).

Table 3 – Repeated ideas: second theme

“ChatGPT is a prototype dialogue-based AI chatbot capable of understanding natural human language
and generating impressively detailed human-like written text.”
“The dialogue format makes it possible for ChatGPT to answer follow-up questions.”
“[…] it is capable of providing descriptions, answers and solutions to complex questions.
“Real-world applications could include generating content for websites, answering customer
inquiries, providing recommendations as well as creating automated chatbots.”
“It’s a tool that is, already […] useful. […] it can answer questions that would be wildly frustrating
to Google, like explaining why a particular joke is funny […]. Use it like a text generator, and it
can create reams of text that are clear, plausible and generally bland […]. You can bounce ideas off
it, request rephrasings, summaries or expansions, and then finesse the work in the finished
project.”
“[…] an upgrade to GPT-3 that allows for complex rhyming poetry.”
“[…] might be able to write a passable job application, school essay or grant application [...]”
“[…] a new AI chatbot that can write convincing essays –clear and coherent responses written in
natural English […] can provide convincing human-sounding answers to exam questions,
The use of AI tools to generate writing that can be passed off as one’s own […]”
“[…] that is capable of writing cogent essays, solving science and math problems and producing
working computer code… Students are using it to write their assignments, passing off AI-generated
essays and problem sets as their own.” “It performs reasonably well across a wide variety of tasks
and academic subjects.”
“ChatGPT: Student builds app to sniff out AI-written essays […]”
“Let me be candid (with apologies to all of my current and former students): What GPT can produce
right now is better than the large majority of writing seen by your average teacher or professor
It also managed to compose a convincing 400-word “friendly” cover letter for an application to be a
manager at Starbucks.”
Theme 2: Writing can be done by a machine.
Source: created by the author (2023).

The repeated ideas in Table 3, expressed by words highlighted in bold and also
represented by the following excerpts: “[...] One academic… described it as writing with the
style and knowledge of a smart 13-year-old... the system also makes a fantastic partner
[...]follow”; “Silicon Valley firm insists its new text generator, which writes human-sounding
essays […]”; “[…] a free online chatbot that can expertly write almost anything, from English
essays and news articles, to meal plans and computer code [...]” exemplify various possible uses
 64

of ChatGPT to generate human-like texts in societal contexts, particularly in education contexts.

By recognizing these repeated ideas, I categorized the second theme, named writing can be
done by a machine.
The third theme revealed in the initial moments of the analysis was categorized as
writer’s block is over. Since these three themes established a connection, I propose the first
theoretical construct: WRITING CAN BE DONE BY “ANY” “ONE”. In addition, four other
themes were revealed. However, up to this moment in my research path, I only established a
connection between two themes: writing essays serves the purpose of assessment in education;
writing essays has become easier and faster, then I elaborate the second theoretical construct:
WRITING IS CORE TO ASSESSMENT IN EDUCATION (see Table 4).

Table 4 – Theoretical constructs and themes

WRITING CAN BE DONE BY “ANY” “ONE”

Theme 1: Writing is impacted by GenAI technologies, especially ChatGPT
Theme 2: Writing can be done by a machine
Theme 3: Writer’s block is over

WRITING IS CORE TO ASSESSMENT IN EDUCATION

Theme 4: Writing essays serves the purpose of assessment in education
Theme 5: Writing essays has become easier and faster

Source: created by the author (2023).

Following the conventions by Auerbach and Silverstein (2003), theoretical constructs

are written in capital letters, revealed themes are written in italics, and repeated ideas are the
terms between quotation marks. In this example, five thematic categories were revealed and
organized into two theoretical constructs. Ultimately, these constructs are discussed in light of
Complex Thinking along with the theoretical bases forming the theoretical narrative.
Conducting research through the lens of Complex Thinking and adopting Analytical
Entwining as a method of data interpretation enables researchers to address epistemological and
methodological implications that arise from the undeniable subject-object interconnectedness.
Moreover, it allows for the expression of a researcher who acknowledges themselves as being
historically and contextually situated with perspectives shaped by societal constructs (Boa Sorte
et al., 2023). This interpretive approach rejects the idea of fixed truths, and understands
 65

knowledge and complex systems as mutable and random implying strategies play a crucial role
in a research journey (Morin, 2015; Petraglia; Moraes, 2022).
Through the lens of Complex Thinking, I describe “screenshots of the reality I see” as a
strategy in my research path. This term refers to the gathered documents used to interpret the
interconnections among the phenomena of writing, education, GenAI technologies, and societal
contexts represented by the phenomena of digital newspapers and website articles, social media
posts, YouTube videos, and podcasts. This term highlights how perceptions of reality are
subjective and multifaceted, including personal experiences, societal constructs, and cognitive
biases. Using the metaphor “screenshots” suggests that an individual’s perception of reality
captures a momentary, partial view, and is visible at a specific time. I understand this aligns
well with the principles of Complex Thinking because it acknowledges that our understanding
of reality is always limited and constructed from specific viewpoints.
Furthermore, among the numerous documents I collected over the four years of this
investigation, 60 were selected to be analyzed. Considering the abundance of data gathered,
which is a major concern for researchers (Freeman, 1988; Auerbach; Silverstein, 2003), I
propose integrating GenAI technology into the Analytical Entwining movements. This
integration serves as the second strategy to engage with data in my research journey.
Building on this, incorporating a GenAI platform represents a strategic approach to
enhance data interpretation, leveraging the capabilities of the ChatGPT/GPT-4 model (OpenAI,
2023c). My collaboration with ChatGPT facilitated the execution of the four analytical
movements across two distinct human-machine interactions. The first interaction was designed
to reach specific objective No. 1: to examine the approach to writing conveyed on internet-
based platforms, particularly after the launch of ChatGPT. The second interaction aimed to
explore educational, pedagogical, and ethical issues related to the impact of AI-text generators,
especially ChatGPT, raised on internet-based platforms. In both interactions, ChatGPT
generated theoretical constructs, assisting in meeting the general objective of this research: to
apprehend the concept of the phenomenon of writing impacted by the adoption of GenAI in
societal contexts in the field of education.
I now introduce Analytical Entwining integrated with GenAI (adapted from Boa Sorte
et al., 2023; dos Santos; Boa Sorte; de Oliveira, forthcoming). This method of analysis requires
careful prompt engineering to maximize the potential of the GenAI platform. I designed four
specific prompts to conduct the analysis, each tailored to address individual characteristics of
the analytical movements required for this method (see Figure 22).
 66

Figure 22 – Analytical Entwining Integrated with GenAI

Source: created by the author based on Boa Sorte et al. (2023).

Unlike data analysis performed individually by a human, ChatGPT can concurrently

process up to ten documents. This LLM is also capable of executing the first two analytical
movements simultaneously. Within the framework of Analytical Entwining integrated with
GenAI, the analysis begins with ChatGPT “locating relevant excerpts” in each document and
“recognizing repeated ideas” across all the documents uploaded by the researcher. This initial
process is driven by a single prompt provided by the researcher at the start of the collaboration
with ChatGPT. Prompt No. 1, originally intended for analyzing textual documents, proved
inadequate for social media posts. Consequently, prompt No. 2 was specifically designed to
code documents saved as images and also to execute the first two movements. Once repeated
ideas have been recognized across all document types, prompt No. 3 is employed to organize
the “macro list of ideas”, laying the groundwork for the theme generation, and “categorizing
revealed themes”. Lastly, prompt No. 4 delivers the essential commands enabling ChatGPT to
establish connections among the “revealed themes” and finalize the analysis by “elaborating
theoretical constructs”.
 67

For this purpose, I utilized the Prompt Creation Reference Chart, which outlines three
key elements crucial for constructing effective prompts: “(1) the prompt’s author: background
information, (2) the main characteristics of the request: design, target audience, and theme, (3)
and the purpose of the prompt: what the prompt is for, what the user wants to achieve” (dos
Santos; Boa Sorte; de Oliveira, forthcoming). This Reference Chart is a principle to promote
the creation of well-structured initial prompts. As outlined by this principle, the request
component — the second element — can be customized according to the specific context, thus
enabling GenAI users to craft their prompts more effectively. To align with the research
requirements, I modified this component by replacing Target Audience with Coding Criteria.
This alteration enables researchers to incorporate specific coding criteria essential for
conducting data analysis (see Figure 23).

Figure 23 – Prompt Creation Reference Chart 1

Source: adapted from dos Santos, Boa Sorte & de Oliveira (Forthcoming).

For ethical considerations, the formulation of prompt No. 1 was developed without
disclosing the research objectives to ChatGPT, guided by the Prompt Creation Reference Chart
1. This first prompt, as detailed in Table 5, serves as the foundational basis for the subsequent
three prompts adopted in Analytical Entwining integrated with GenAI.
 68

Table 5 – Visualizing Prompt No. 1

I am a PhD candidate researching in the area of education, languages, and technologies. I am currently
focusing on coding the collected data. Generate a list of excerpts that address the theme of research, a list
of the repeated ideas based on these excerpts, and a table format, connecting the repeated ideas with each
excerpt. The research focuses on generative artificial intelligence in education, AI-text generator
platforms, and writing. This task needs to follow some criteria: the excerpts must come from the documents
I uploaded, the excerpts selected must retain the exact language used by the authors of the documents,
the list of the repeated ideas must be based on the excerpts, and the excerpts selected must align with the
research theme. Generate these lists bearing in mind the purposes: to locate relevant excerpts that address
the theme of this research, and to recognize repeated ideas from those excerpts.

AUTHOR (WHO): BACKGROUND INFORMATION

REQUEST (MAIN CHARACTERISTICS): DESIGN – THEME – CODING CRITERIA

PURPOSE (WHAT FOR): PURPOSE 1 + PURPOSE 2

Source: created by the author (2023).

It is important to note that the background information — represented as the “author”

component — and the “theme”, which forms part of the “request” component, remain the same
across the four prompts created for this data analysis. The choice of language, which informs
the “theme” and the other main characteristics of the prompt, is critical. It provides the LLM
with sufficient detailed information, enabling the GenAI platform to generate the required
outputs without having to inform the research objectives.
To demonstrate the significance of language choice in prompt engineering, I revisit the
rationale for customizing a second prompt that enables ChatGPT to execute the analytical
movements of “locating relevant excerpts” and “recognizing repeated ideas”. During tests with
ChatGPT, it became apparent that prompt No. 1 was unable to process image documents due
to the use of the term “excerpt”. Consequently, for image data, the term “passage” was
employed. This led to the refinement of the initial prompt, in accordance with Prompt Creation
Reference Chart 1, resulting in the development of prompt No. 2 (see Figure 24).
 69

Figure 24 – Prompt Creation Reference Chart 2

Source: adapted from dos Santos, Boa Sorte & de Oliveira (Forthcoming).

Continuing as a component of the “request” element, the “design” is distinguished as a

key feature of the prompt, encompassing the structuring of text, visuals, and additional forms.
The design delineates the manner in which ChatGPT is expected to arrange the format of the
response. This component is essential to specify how ChatGPT should configure the layout of
the output. The layouts of this method of analysis include lists and tables. The “design”
component of prompts No. 1 and No. 2, as illustrated in Tables 5 and 6, requests ChatGPT to
deliver a ‘list of repeated ideas’ based on a list of ‘relevant information’ that addresses the
outlined theme of research and a table connecting the repeated ideas with this selected ‘relevant
information’.
 70

Table 6 – Visualizing Prompt No. 2

I am a PhD candidate researching in the area of education, languages, and technologies. I am currently
focusing on coding the collected data. Generate a list of passages that address the theme of research, a list
of the repeated ideas based on these passages, and a table format, connecting the repeated ideas with each
passage. The research focuses on generative artificial intelligence in education, AI-text generator
platforms, and writing. This task needs to follow some criteria: the passages must come from the images I
uploaded, the passages selected must retain the exact language used by the authors of the documents, the
list of the repeated ideas must be based on the passages, and the passages selected must align with the
research theme. Generate these lists bearing in mind the purposes: to locate relevant passages that address
the theme of this research, and to recognize repeated ideas from those passages.

AUTHOR (WHO): BACKGROUND INFORMATION

REQUEST (MAIN CHARACTERISTICS): DESIGN – THEME – CODING CRITERIA

PURPOSE (WHAT FOR): PURPOSE 1 + PURPOSE 2

Source: created by the author (2023).

The “design” element influences the perception and comprehension of information.

Well-designed text layouts enhance communication effectiveness by simplifying the analysis
process. These “design” principles were integral during my collaboration with ChatGPT.
During the first two movements of analysis, ChatGPT successfully “located relevant
information” across various document types and “recognized repeated ideas”. The results were
organized into two lists and one table, illustrating the connections between them as specified in
the “design” requirements of prompts No. 1 and No. 2, as demonstrated in Table 7.
 71

Table 7 – Visualizing Output Text Document 4C

ChatGPT

From the document titled "Should we let students use ChatGPT? TED Talk," here's a structured analysis for
your research:
Excerpts Addressing the Theme

1. Excerpt 1: "This TED Talk discusses the integration of AI tools like ChatGPT in educational
settings and debates whether their use should be encouraged or restricted."(#4C Should we let
stude…)
2. Excerpt 2: "The presenter argues that while AI can significantly enhance learning, it also raises
concerns about academic integrity and the development of critical thinking skills."(#4C Should we
let stude…)
3. Excerpt 3: "Examples are given of schools that have successfully integrated ChatGPT to improve
engagement and personalize learning, highlighting positive outcomes."(#4C Should we let stude…)
4. Excerpt 4: "Concerns are also voiced about students becoming overly reliant on AI for homework
and assignments, potentially hindering their learning process."(#4C Should we let stude…)
Excerpt 5: "The talk concludes with a call for educational institutions to develop clear guidelines and policies
that maximize the benefits of AI while mitigating its risks."(#4C Should we let stude…)
Repeated Ideas Based on Excerpts

● AI Integration in Education: Discusses the incorporation of AI tools like ChatGPT in educational

practices, focusing on both the positive impacts and challenges.
● Academic Integrity and Dependency Concerns: Emphasizes the potential risks of AI in education,
particularly in terms of academic integrity and student dependency
Policy Development for AI Use: Advocates for the development of structured policies and guidelines to
manage the use of AI in educational settings effectively.
Table Connecting Repeated Ideas with Excerpts

Repeated Idea Connected Excerpt

AI Integration in Education Excerpt 1, 3
Academic Integrity and Dependency Concerns Excerpt 2, 4
Policy Development for AI Use Excerpt 5

This analysis provides insights into the complex dynamics of using AI like ChatGPT in educational contexts,
exploring the balance between leveraging technological advancements for enhanced learning and ensuring
that students remain the primary agents in their educational journeys.
Source: created by the author in collaboration with ChatGPT (OpenAI, 2023c).

This table displays the responses of the ChatGPT/GPT-4 model (OpenAI, 2023c) to
prompt No. 1 after completing the first two analytical movements on text document 4C, a
YouTube video transcript. Initially, the model generates a list of five excerpts that address the
outlined theme of this study. Then, it produces a second list including the three ideas recurring
within these excerpts. The second analytical movement concludes with a table where each row
connects a thematic idea with its corresponding excerpts from the document.
 72

Once all document types are individually coded, the analysis can proceed to the third
analytical movement: “categorizing revealed themes”. This phase involves grouping recurring
ideas into coherent categories. Such thematic organization of data is supported by the “macro
list of ideas”, which not only facilitates this process but is also indispensable for laying the
foundational groundwork necessary for generating themes. Although this procedure is part of
the “recognizing repeated ideas” movement, it cannot be executed in isolation during prompts
No. 1 and No. 2. This is because it involves compiling a list of repeated ideas identified across
all document types, texts and images. Consequently, the organization of a “macro list of ideas”
is incorporated into the “design” element in prompt No. 3 as specified in the Prompt Creation
Reference Chart 3 (see Figure 25).
 73

Figure 25 – Prompt Creation Reference Chart 3

Source: adapted from dos Santos, Boa Sorte & de Oliveira (Forthcoming).

The final key component of the “design” element tailored for the Analytical Entwining
integrated with GenAI is the “coding criteria”. Prompt Creation Reference Chart 3 outlines how
ChatGPT should implement the “categorizing revealed themes” movement in accordance with
Analytical Entwining, a foundation for the method of analysis. This chart also specifies the
language to optimize responses from the LLM. Language choice in prompt engineering
 74

represents a major and challenging requirement, as demonstrated in prompt No. 3 (see Table
8).

Table 8 – Visualizing Prompt No. 3

I am a PhD candidate researching in the area of education, languages, and technologies. I am currently
focusing on coding the collected data. Generate a ‘macro list of ideas’, a comprehensive list of the recurring
ideas identified across the various lists of repeated ideas from all the documents, and a table titled
‘revealed themes’ grouping the repeated ideas into coherent categories. The research focuses on
generative artificial intelligence in education, AI-text generator platforms, and writing. This task needs to
follow some criteria: the ‘macro list of ideas’ represents a comprehensive list of the recurring ideas across
the lists of the repeated ideas from all the previously uploaded documents, each idea of the ‘macro list of
ideas’ must be expressed as a clause, the vocabulary used to commence each idea from the ‘macro list of
ideas’ must preferably begin with the term ‘writing’ or, alternatively, with any derivative of its root, such
as ‘writer’ or ‘written’, either the term ‘writing’ or any derivative of its root must integrate the subject
of the clause, the ‘macro list of ideas’ forms the basis for the revealed themes, the themes originate from
grouping these repeated ideas into coherent categories, the revealed themes represent conceptual
similarities among these repeated ideas, the vocabulary for each theme must avoid jargon and must use
titles that clearly express the union of the repeated ideas for easy understanding, each ‘theme’ must be
expressed as a clause, the vocabulary used to commence each theme must preferably begin with the term
‘writing’ or, alternatively, with any derivative of its root, such as ‘writer’ or ‘written’, either the term
‘writing’ or any derivative of its root must integrate the subject of the clause, and the themes must align
with the research theme. Generate these lists bearing in mind the purposes: to organize a ‘macro list of ideas’
repeated across the documents , and to categorize ‘revealed themes’ from the macro list of ideas.

AUTHOR (WHO): BACKGROUND INFORMATION

REQUEST (MAIN CHARACTERISTICS): DESIGN – THEME – CODING CRITERIA

PURPOSE (WHAT FOR): PURPOSE 1 + PURPOSE 2

Source: created by the author (2023).

The guidelines are developed to help structure and articulate ideas and themes derived
from previously analyzed documents. They emphasize the consistent employment of specific
vocabulary, especially derivatives of “writing”, to initiate each idea and theme. Furthermore,
the guidelines advocate for the use of clear, jargon-free language and ensure that these themes
align directly with the main research theme, promoting coherence and ease of comprehension.
After collaborating with ChatGPT using prompt No. 3, the model successfully categorized four
themes. The results were organized into two lists and one table, illustrating the connections
between the “revealed themes” and the “macro list of ideas” (see Table 9).
 75

Table 9 – Visualizing Output - Prompt No. 3 - Specific Objective 1

ChatGPT

To align with your research on generative artificial intelligence in education, AI-text generator platforms, and
writing, I will create a ‘macro list of ideas’ and group these ideas into ‘revealed themes’. Each entry will be
crafted as a clause with ‘writing’ or its derivatives as the subject, adhering to the specified criteria to ensure
clarity and relevance to the overarching research theme.

Macro List of Ideas

1. Writing processes are being expedited by AI's ability to automate routine tasks.
2. Writing’s creative dimensions are expanded by AI through offering novel suggestions and
ideas.
3. Writing integrity is being safeguarded by AI tools that ensure ethical practices such as avoiding
plagiarism.
4. Writing tools are becoming more interactive and accessible due to AI’s integration.
5. Writing’s appeal and engagement are enhanced through AI-facilitated community platforms
that foster collaboration.
6. Writing education is evolving as AI provides personalized learning experiences.
7. Writing’s global reach is extended by AI’s multilingual capabilities, making content accessible
across language barriers.

Revealed Themes

1. Writing is enhanced through AI by expediting processes and expanding creative dimensions.

○ Connected Ideas: Writing processes are being expedited by AI; Writing’s creative
dimensions are expanded by AI.
2. Writing integrity is safeguarded by AI tools ensuring ethical practices.
○ Connected Ideas: Writing integrity is being safeguarded by AI tools.
3. Writing becomes more interactive and accessible through AI technologies.
○ Connected Ideas: Writing tools are becoming more interactive; Writing’s appeal is
enhanced through AI-facilitated community platforms.
4. Writing education evolves with AI providing personalized learning experiences and extending
global reach.
○ Connected Ideas: Writing education is evolving as AI provides personalized experiences;
Writing’s global reach is extended by AI’s capabilities.

Table Connecting Macro Ideas with Revealed Themes

Macro Idea Revealed Theme

Writing processes are being Writing is enhanced through AI by expediting processes and expanding
expedited by AI creative dimensions.

Writing’s creative dimensions are Writing is enhanced through AI by expediting processes and expanding
expanded by AI creative dimensions.

Writing integrity is being Writing integrity is safeguarded by AI tools ensuring ethical practices.
safeguarded by AI tools

Writing tools are becoming more Writing becomes more interactive and accessible through AI
interactive technologies.
 76

Writing’s appeal is enhanced Writing becomes more interactive and accessible through AI
through AI platforms technologies.

Writing education is evolving with Writing education evolves with AI providing personalized learning
AI experiences and extending global reach.

Writing’s global reach is extended Writing education evolves with AI providing personalized learning
by AI experiences and extending global reach.

These themes provide a structured overview of how AI impacts writing, aligning with your research focus on
generative AI in education, AI-text generator platforms, and writing. This approach ensures clarity and
cohesiveness in presenting your findings.
Source: created by the author in collaboration with ChatGPT (OpenAI, 2023c).

This table presents the responses of the ChatGPT/GPT-4o model (OpenAI, 2023c) to
prompt No. 3, after analyzing 30 documents selected to achieve specific objective 1. The model
generates a ‘macro list’ of seven recurring ideas from these text and image documents.
Following this, ChatGPT categorized these ideas into four themes, listing them and identifying
each connected idea. The third analytical movement culminates in a table where each row
connects a revealed theme with its corresponding macro ideas from the documents. My
collaboration with ChatGPT concludes in the fourth movement: “elaborating theoretical
constructs”. In this phase, ChatGPT connects the “revealed themes” and generates these
constructs via prompt No. 4, as guided by Prompt Creation Reference Chart 4 (see Figure 26).
 77

Figure 26 – Prompt Creation Reference Chart 4

Source: adapted from dos Santos, Boa Sorte & de Oliveira (Forthcoming).

With Prompt Creation Reference Chart 4, I present the final key element of prompt
engineering in Analytical Entwining integrated with GenAI. This component refers to the
“purpose” of the prompt, which details “what the prompt is for” and “what the user wants to
achieve”; note that this may vary in number. In this study, all the “purposes” inform each
analytical movement in every prompt and are described using consistent terminology that
matches the analytical movements, as demonstrated in prompt No. 4 (see Table 10).
 78

Table 10 – Visualizing Prompt No. 4

I am a PhD candidate researching in the area of education, languages, and technologies. I am currently
focusing on coding the collected [Link] a list of the ‘theoretical constructs’, grouping the ‘revealed
themes’ into coherent categories and a table format, connecting the grouped themes with each
[Link] research focuses on generative artificial intelligence in education, AI-text generator
platforms, and writing. This task needs to follow some criteria: the ‘theoretical constructs’ represent the
conclusion of the coding, the ‘theoretical constructs’ originate from grouping the themes into coherent
categories, the ‘theoretical constructs’ represent conceptual similarities among the ‘revealed themes’,
each ‘theoretical construct’ must be expressed as a clause, the vocabulary used to commence each
‘theoretical construct’ must preferably begin with the term ‘writing’ or, alternatively, with any derivative
of its root, such as ‘writer’ or ‘written’, either the term ‘writing’ or any derivative of its root must
integrate the subject of the clause, and these ‘theoretical constructs’ must align with the research theme.
Generate these lists bearing in mind the purpose: to elaborate ‘theoretical constructs’ establishing
connections among the ‘revealed themes’.

AUTHOR (WHO): BACKGROUND INFORMATION

REQUEST (MAIN CHARACTERISTICS): DESIGN – THEME – CODING CRITERIA

PURPOSE (WHAT FOR): PURPOSE 1 + PURPOSE 2

Source: created by the author (2023).

Prompt No. 4 also exemplifies the challenges and importance of language choice in
“coding criteria”, as previously discussed in prompt No. 3. The guidelines outline methods for
structuring theoretical constructs that summarize research findings. They highlight the
consistent use of specific terminology related to “writing” to initiate each construct and require
that these constructs be clearly phrased and aligned with the main research theme. The
constructs themselves are developed by grouping related themes, thus emphasizing their
conceptual links, as demonstrated in Table 11.

Table 11 – Visualizing Output - Prompt No. 4 - Specific Objective 1

ChatGPT

To create theoretical constructs that encapsulate the revealed themes from your research, we will group the
themes into coherent categories, ensuring each construct starts with "writing" or its derivatives as the subject.
These constructs will summarize the findings and establish clear connections among the themes, aligning with
your research focus on generative artificial intelligence in education, AI-text generator platforms, and writing.

Revealed Themes Recap

1. Writing is enhanced through AI by expediting processes and expanding creative dimensions.

2. Writing integrity is safeguarded by AI tools ensuring ethical practices.
3. Writing becomes more interactive and accessible through AI technologies.
4. Writing education evolves with AI providing personalized learning experiences and extending
global reach.
 79

Theoretical Constructs

1. Writing is revolutionized by AI through enhancement and accessibility.

○ Grouped Themes: Writing is enhanced through AI by expediting processes and expanding
creative dimensions; Writing becomes more interactive and accessible through AI
technologies.
○ Description: This construct combines the themes of enhancement and accessibility,
highlighting how AI transforms the writing process by increasing efficiency, creativity, and
user engagement.
2. Writing integrity is maintained through AI-driven ethical safeguards.
○ Grouped Themes: Writing integrity is safeguarded by AI tools ensuring ethical practices.
○ Description: This construct focuses on the ethical aspects of AI in writing, emphasizing the
role of AI tools in preserving the integrity and authenticity of written content.
3. Writing education is transformed by AI through personalization and global reach.
○ Grouped Themes: Writing education evolves with AI providing personalized learning
experiences and extending global reach.
○ Description: This construct addresses the impact of AI on education, particularly how AI
facilitates personalized learning and broadens the accessibility of educational content across
languages and regions.

Table Connecting Grouped Themes with Theoretical Constructs

Theoretical Construct Grouped Themes

Writing is revolutionized by AI through Writing is enhanced through AI by expediting processes

enhancement and accessibility. and expanding creative dimensions; Writing becomes
more interactive and accessible through AI technologies.

Writing integrity is maintained through AI-driven Writing integrity is safeguarded by AI tools ensuring
ethical safeguards. ethical practices.

Writing education is transformed by AI through Writing education evolves with AI providing

personalization and global reach. personalized learning experiences and extending global
reach.

Source: created by the author in collaboration with ChatGPT (OpenAI, 2023c).

This final table displays the responses of the ChatGPT/GPT-4o model (OpenAI, 2023c)
to prompt No. 4, following the completion of the fourth analytical movement, “elaborating
theoretical constructs”. Initially, the model revisits the “revealed themes”. Although not
required by the “design” element of Prompt Creation Reference Chart 4, this repetition aids the
researcher in monitoring the process. Subsequently, the model generates a list of three
theoretical constructs, specifying which themes were grouped together to form each construct
and providing a brief description of each. The last analytical movement concludes with a table
where each row links a theoretical construct to its corresponding revealed themes from the
documents.
 80

In studies in which the research data contains only one document type, the researcher
may choose either prompt No. 1 for text documents or prompt No. 2 for image documents. This
choice is based on the fact that both prompts encompass the initial and subsequent analytical
movements of Analytical Entwining integrated with GenAI.
It is noteworthy that ChatGPT has the capacity to concurrently handle a maximum of
ten documents. Nonetheless, a deliberate choice was made to adhere to the principles of
Analytical Entwining, which serve as the foundational framework for this analytical method.
Consequently, commands were issued to ChatGPT and documents were sequentially uploaded.
The interaction with the model unfolded in two distinct phases, each tailored to each
specific objective. Initially, text document 1A was uploaded and ChatGPT was instructed with
prompt No. 1 until completion of this document type, followed by the commencement of work
on each image document type. Subsequently, image document 2A was uploaded, and ChatGPT
was directed with prompt No. 2 until completion of this document category. This
methodological procedure was adopted after preliminary tests with the model, which
demonstrated the efficacy of overseeing the coding process and evaluating ChatGPT’s
responses in alignment with the coding criteria of Analytical Entwining integrated with GenAI.
Despite its enhanced capabilities, GPT-4 shares certain limitations with its predecessors,
such as the potential to “hallucinate” facts and commit reasoning errors (OpenAI, 2023a). In
this research, continuous monitoring was imperative due to the occurrence of “hallucinations”
or erroneous outputs during the interactions with ChatGPT. “Hallucinations”, in the context of
LLMs, refer to instances where the model generates factually incorrect or misleading
information presented as truth, which can compromise the integrity of data processing and
output. Therefore, vigilant oversight during human-AI collaboration is essential to ensure the
ethical, efficient, and responsible use of these AI text generators. This proactive approach
facilitates the mitigation of risks associated with AI-generated content, reinforcing the
importance of human oversight in managing and correcting AI behavior, encompassing
strategies from rigorous human review to providing additional contextual grounding for
responses.
In this study, the 60 documents analyzed were selected based on the following inclusion
criteria: the trajectory of technological advancements in ChatGPT and this research specific
objectives. Regarding ChatGPT, I considered five intervals that refer to the time between events
or milestones defined by specific technological releases of OpenAI’s LLMs (see Figure 27).
 81

Figure 27 – The Journey of OpenAI GPT Models

Source: created by the author based on Open AI (2015-2024).

The initial interval is associated with the release of GPT-3, representing a milestone in
the field of NLP owing to its unprecedented capacity in generating human-like text through
extensive pre-training. The subsequent interval begins with the launch of ChatGPT, a
groundbreaking variant of GPT-3 known as GPT-3.5. This version introduced enhanced
conversational capabilities, allowing for more coherent and context-aware interactions.
Additionally, this period is notable for ChatGPT reaching 100 million users within two months,
highlighting its widespread acceptance and applicability.
The third interval starts on March 14, 2023, with the public release of GPT-4. This
model has introduced noticeable improvements in text comprehension and generation,
delivering more precise and sophisticated responses. GPT-4 also marks the beginning of the
fourth interval on September 25, 2023, with OpenAI’s announcement of the model’s
multimodal capabilities. These developments signal a transition to integrating voice and image
capabilities, enhancing the interface to support vocal interactions and enabling users to provide
visual inputs to ChatGPT for more intuitive communication.
 82

The fifth and final interval begins on May 13, 2024, with the introduction of GPT-4o,
showcasing the latest advancements in deep learning to date. The letter “o” stands for omni —
suggesting OpenAI’s intention to achieve omnipresence — with capabilities such as human-
like response times in conversations and significant improvements in handling non-English
language text. These intervals compose the “screenshots of the reality I see”, defined in
accordance with the Complex Thinking epistemology.
Concerning the second inclusion criterion, documents were selected based on their
relevance to each specific objective. To achieve the first objective — to examine the approach
to writing conveyed on internet-based platforms, particularly after the launch of ChatGPT —
the first 30 documents focusing on GenAI and writing in educational contexts were selected.
Additionally, to address the second specific objective, another 30 other documents were chosen
for data analysis. The selection was influenced by the three main contexts: pedagogical
(associated with methods or theory of teaching writing), ethical (concerning beliefs, what is
morally right or wrong), and educational issues (related to the impact of AI text generators,
especially ChatGPT, as discussed internet-based platforms). These criteria are depicted in
Figure 28, the final image of this section, in which I also summarize my methodological path.
 83

Figure 28 – Methodological Path

Source: created by the author based on Boa Sorte et al. (2023), Morin (2015), and Petraglia and
Moraes (2022).
 84

This image depicts the two strategies implemented to achieve my research objectives.
The first strategy, “screenshots of the reality I see”, consists of screenshots numbered from 1 to
5. These numbers indicate predefined intervals representing the technological advancement
trajectory of ChatGPT. Each screenshot is labeled with a combination of a letter and a number
and contains two documents. Documents labeled with letters A, B, and C relate to specific
objective number one and the holographic principle, while letters D, E, and F align with specific
objective number two. This second objective engages with the three principles of Complex
Thinking: recursive, holographic, and dialogic. The “screenshots of the reality I see” symbolize
the interconnectedness of the phenomena investigated in this study: writing, education, GenAI
technologies, and the principles of Complex Thinking. The second strategy integrates
ChatGPT’s capabilities into the four analytical movements via prompt engineering. A collection
of 60 documents, including Instagram posts, website articles, digital newspaper articles,
podcasts, and YouTube videos, was analyzed using Analytical Entwining integrated with
GenAI. This analysis resulted in the generation of eight theoretical constructs, which I discuss
in the subsequent two sections.
 85

SECTION 3

APPROACHES TO WRITING IN EDUCATION

WITH GENERATIVE AI10

10
Image generated using OpenAI’s DALL-E image generation tool on June 28, 2024. The prompt used for
generation was: “I am a PhD candidate researching in the area of education, languages, and technologies. I am
currently focusing on coding the collected data. Generate an image that depicts ChatGPT working on data
analysis. The research focuses on generative artificial intelligence in education, AI-text generator platforms,
and writing. This task needs to follow some criteria: the images must show the data: Instagram posts, Digital
newspaper articles, Podcasts, and YouTube videos, the image must align with the research theme. Generate an
image bearing in mind the purposes: to showcase a representation of analytical movements carried by
ChatGPT”.
 86

3 APPROACHES TO WRITING IN EDUCATION WITH GENERATIVE AI

“Writing has been with us now for so long

that we’ve lost the habit of thinking of it
as an invention” (Dibbell, 2010, p. 1).

This section opens with a summary of the four themes that were essential in developing
the constructs discussed here (see Figure 29) 11. In this analytical path, considering the
“screenshots of the reality I see” in light of Complex Thinking principles (Morin, 2015)
symbolize the interconnectedness of the phenomena investigated in this research: writing,
education, and GenAI technologies.

11
Following the conventions by Auerbach and Silverstein (2003), theoretical constructs are written in capital
letters and revealed themes are written in italics.
 87

Figure 29 – Theoretical Constructs: specific objective 1

Source: created by the author (2023).

 88

I initiate the presentation and analysis of the research findings, directly addressing the
first specific objective of the study. It refers to writing as a phenomenon, given that it is a life
experience (Ravagnoli, 2017), interconnected with societal aspects, and recognizes writing as
an inherently social practice. This understanding is grounded in the holographic principle of
Complex Thinking (Morin, 2015), which acknowledges the interrelatedness and
interdependence of all components within a system. The documents analyzed in this study —
YouTube videos, podcasts, website and digital newspaper articles, and social media posts —
serve as bridges, capturing “screenshots of the reality I see” that exemplify this
interconnectedness.
Following this principle, every component of the writing process is embedded within a
larger societal context, reinforcing that no element stands alone. The interpretation of words
and texts depends heavily on their contextual use, underscoring the necessity to view them
within their societal, cultural, and knowledge frameworks. By employing a cognitive
framework that fosters connections among seemingly unrelated elements, writing studies are
positioned as integral to broader educational and societal contexts. Thus, education must adapt
to societal and technological changes that impact writing, reflecting the holographic principle
where each individual embodies societal aspects through language, culture, and norms (Morin,
2000).
This section is divided into three subsections, each focusing on a different theoretical
construct elaborated through the movements of Analytical Entwining integrated with GenAI
and in collaboration with ChatGPT. For each construct, the discussion begins by outlining its
foundational themes. Subsequent analysis involves citing scholars who comment on these
constructs, using excerpts from the literature and research data to deepen the discussion. This
approach is systematically applied to all three constructs.

3.1 Writing is revolutionized by AI through enhancement and accessibility

The theoretical construct WRITING IS REVOLUTIONIZED BY AI THROUGH

ENHANCEMENT AND ACCESSIBILITY12 combines themes that fundamentally alter the
nature of writing. Specifically, GenAI enhances writing by expediting processes and expanding
creative dimensions, simultaneously making writing more interactive and accessible through

12
Following the conventions by Auerbach and Silverstein (2003), theoretical constructs are written in capital
letters.
 89

advanced technologies. This construct emphasizes the transformation of the writing process
through increased efficiency, creativity, and enhanced user engagement. The merging of these
themes underscores a shift towards a more dynamic and accessible writing environment
facilitated by AI innovations.
Building on the ways GenAI fundamentally alters the nature of writing, examining this
construct through the lens of Complex Thinking principles (Morin, 2015) offers deeper insight
into the systemic changes within this domain. AI text generators continuously refine and evolve
through iterative feedback loops from user interactions, demonstrating the recursive principle
by reflecting each layer of improvement back onto the system, leading to progressively
sophisticated writing aids. Furthermore, each AI tool, while part of a larger system,
encapsulates the entire functionality of the writing process but contributes uniquely to specific
aspects like grammar enhancement and style personalization.
This encapsulation demonstrates the holographic principle, as every component
contains the whole in its functionality, mirroring the entire writing ecosystem. This refers to
Morin’s view, in which the whole is greater than the sum of its parts because each part contains
specific characteristics that are not apparent when considered only as part of the whole. The
ongoing interactions between these AI tools and the users create a dialogic relationship,
fostering co-evolutionary development in writing practices and illustrating the dialogic
principle. This complex interplay of components reflects the system as a whole and highlights
how each part contributes to and is influenced by the system, underscoring the notion of system.
Together, these principles illustrate a sophisticated convergence of technology and human
creativity, reshaping the landscape of written communication.
Technological advancements have historically reshaped the nature of writing. The
transition from paper pages to digital screens signifies a revolutionary change, driven by the
continuous development of technological devices (Kist, 2005). Further emphasizing this shift,
the profound disruption of the Internet has altered several elements of society (Christensen,
2012), including the nature of writing, particularly its increasing use for online communication
(Leu et al., 2016). Both in research and classroom settings, the implications of these changes
regarding the nature of writing in an online world, though still emerging, are being reconsidered
and expanded (Christensen, 2012; Kist, 2005; Leu et al., 2016). Consequently, a redefinition of
texts and practices is proposed. This involves a pedagogical approach in which diverse forms
of communication and meaning-making—visual, auditory, spatial, behavioral, and gestural
modes—are acknowledged. It represents a transition from literacy (in the singular) to literacies
(in the plural) (Kalantzis; Cope; Pinheiro, 2020).
 90

The creation of writing, once a groundbreaking development, has become so embedded

in human culture and social practices that we often forget its innovative beginnings. As
mentioned in this section epigraph, writing’s longstanding presence has diminished our sense
of its original impact (Dibbell, 2010). However, this perspective is being renewed with the
emergence of AI text generators, which represent a new wave of transformative technology in
written communication. The application of GenAI technologies such as ChatGPT in writing is
evidenced through specific prompts that catalyze both the speed and creativity of the writing
process as shown in the screenshot 4C (see Figure 30).

Figure 30 – “Screenshot of the reality I see” 4C

Source: @eldessaullo (2023).

This image demonstrates the transformative power of GenAI in writing, highlighting

how descriptive prompts used during collaboration with LLMs, like ChatGPT, can enhance
both efficiency and creative output. It showcases how these prompts accelerate the writing
process and expand creative possibilities, aiding authors in tasks from ideation to designing
compelling book covers.
These GenAI systems are altering how we create and engage with text, reminiscent of
how the original invention of writing revolutionized human communication by allowing the
recording and sharing of knowledge across distances and generations (Fischer, 2021). By
automating the writing process, AI text generators are broadening the horizons of written
 91

expression and compelling us to rethink the concepts of authorship and creativity. This ongoing
evolution highlights that writing is not a static achievement but continues to change
dynamically in response to technological advancements. Adopting the term “evolution” implies
understanding that it signifies “not quality grades, nor [...] stages in a model of ‘writing
evolution’ (which does not exist). They are simply different forms that accommodate different
linguistic and social needs as they arise” (Fischer, 2021, p. 329).
GenAI substantially improves the writing process by enhancing efficiency and
broadening creative aspects through a variety of innovative applications. GenAI-driven tools
like ChatGPT and other LLMs streamline the writing process by reducing the time required to
complete tasks and improving the quality of output. For instance, in a study involving midlevel
professional writing tasks, the use of ChatGPT decreased the average time taken by 40% and
increased output quality by 18%, demonstrating major productivity gains and enjoyment of the
writing process for users (Noy; Zhang, 2023). The screenshot 3B (see Figure 31) also illustrates
the advancements in AI-powered content creation, as ChatGPT enables the creation of diverse
digital content in a matter of seconds, reshaping user interactions across multiple platforms.

Figure 31 – “Screenshot of the reality I see” 3B

Source: @AI Chatbot (2023).

 92

The illustration offers a glimpse into how GenAI, particularly through technologies like
ChatGPT, enhances content production efficiency. This is evidenced by its capacity to produce
articles, social media updates, and various other forms of digital content rapidly and effectively.
Users’ comments highlighting the positive impact and growing reliance on GenAI applications
reflect the enthusiasm for new technology and its practical benefits in daily digital interactions.
However, these views contrast with many opposing and skeptical opinions arguing many
responses. This duality aligns with a historical narrative of responses, where new technology is
often met with either enchantment or fear (Lemos, 2013). This often leads to neglect ethical
concerns, dependency issues, and the potential decline in critical thinking due to excessive
reliance on AI.
In examining these phenomena, the principles of Complex thinking (Morin, 2015)
emerge as a dynamic network where elements interact in intricate and interdependent ways,
leading to the emergence of new properties. I first identify the holographic principle because
the influence of GenAI on content production reflects the intricate whole within each part,
where each instance of GenAI use encapsulates the broader system’s characteristics. Secondly,
the recursive principle is expressed by the efficacy of GenAI technologies like ChatGPT
involving repeated cycles of feedback and adaptation, where outcomes inform ongoing
improvements and user behaviors.
Thirdly, the dialogic principle emerges through the continuous interaction and
negotiation between human users and GenAI systems, fostering a dynamic interplay that shapes
content creation and digital practices. The dialogic principle is also highlighted by the
contrasting views within the community that uses AI text generators. The enthusiastic embrace
of these platforms for practical benefits is in constant dialogue with skeptical views that caution
against potential downsides, such as ethical neglect and reduced critical thinking. This principle
emphasizes the importance of ongoing conversations and multiple viewpoints that shape the
understanding and evolution of AI technologies.
Lastly, the notion of system is crucial, recognizing that GenAI operates within a
complex, interconnected network of technologies, users, and educational practices, each
influencing and being influenced by the other. These principles state the need for an integrative
approach to understanding and responding to the complexities introduced by GenAI in content
production.
Similarly, LLMs utilized in scientific writing possess the capability to efficiently
produce text of superior quality at a rapid pace. It is crucial, however, to carefully consider
ethical aspects, including the impact on employment opportunities and the risk of disseminating
 93

inaccurate or deceptive information, and take appropriate measures to mitigate these concerns
(Kammer et al., 2023).
Beyond efficiency, GenAI also extends the creative dimensions of writing. For example,
AI has the capability to create new compositions from existing images and aid artists in
transcending the constraints of the human mind, thereby stimulating novel forms of creativity
and innovative designs (Dawood, 2023). In creative writing, AI’s NLP abilities can propose
new content and ideas, cultivating a “machine-in-the-loop” strategy that assists writers in
discovering new creative pathways, despite potential variations in its impact on creativity
among users (Plate; Hutson, 2022). Additionally, the emergence of writing robots capable of
managing the entire content creation process — from selection to publication — showcases
AI’s potential to industrialize and enhance article production, significantly elevating creative
efficiency (Li; Zhang, 2020).
GenAI reshapes creative practices by acting as a filter and gatekeeper, integrating
diverse information sources and mediating between human activity and these sources, which
can lead to the development of novel ideas, but reduce the qualitative diversity of inspiration
(Atkinson; Barker, 2023). These discussions, highlighted in the screenshot 5B (see Figure 32),
emphasize the popular desire for AI to serve as an assistance in mundane chores, ensuring that
humans can reserve their time and mental energy for more creative pursuits like art and writing.
 94

Figure 32 – “Screenshot of the reality I see” 5B

Source: @[Link] (2024).

The image captures key sentiments from social media, emphasizing the role of AI as a
supportive tool rather than a replacement for human creativity. It reflects a broader societal
conversation on preserving human-centric tasks, particularly those involving creativity, while
allowing AI to handle routine tasks like dishwashing and laundry. This distinction clarifies
public expectations and envisions a future where GenAI improves our daily lives by taking over
ordinary tasks, enabling deeper involvement in rewarding and creative activities.
The integration of GenAI in writing also extends to addressing writer’s block, where
LLMs can serve as a conversational partner, mimicking the benefits of talking to others, helping
writers navigate through creative blocks, and maintaining the writer’s authentic voice (Gilburt,
2023; Jacob; Tate, 2023).
In the light of Morin’s (2015) Complex Thinking, in this dynamic network, AI interacts
with various elements of the writing process—such as routine tasks, creative methods, and
content improvement strategies—in intricate and interdependent ways. This interaction fosters
the emergence of new properties and behaviors, including enhanced efficiency and expanded
creative potential, thereby extending the boundaries of human creativity (Dawood, 2023;
 95

Gilburt, 2023; Jacob; Tate, 2023; Li; Zhang, 2020; Noy; Zhang, 2023; Plate; Hutson, 2022;
Reza et al., 2023).

3.2 Writing integrity is maintained through AI-driven ethical safeguards

In the domain of ethical practices within writing, the construct WRITING INTEGRITY
IS MAINTAINED THROUGH AI-DRIVEN ETHICAL SAFEGUARDS is supported by
themes asserting that writing integrity is safeguarded by AI tools that ensure adherence to
ethical standards. It highlights the critical role of AI technologies in upholding the integrity and
authenticity of written content. By focusing on the ethical dimensions of AI applications in
writing, this construct stresses the importance of maintaining truthful and reliable content,
which is fundamental in preserving public trust and the credibility of written works.
The integration of AI tools in academic environments has divided opinions among
educators and scholars. “Ci sono due campi nel mondo accademico. Il primo è quello degli
‘early adopters’ dell’intelligenza artificiale, il secondo è quello dei professori e degli
accademici che pensano che l’AI corrompa l’integrità accademica” (my translation). This
excerpt from the screenshot 3C (see Figure 33) captures the core debate: the first group consists
of early adopters who embrace AI technologies such as ChatGPT for their potential to enhance
research and educational practices, while the second group includes educators and scholars who
view the widespread adoption of AI tools as a potential threat to the established standards of
academic integrity.
 96

Figure 33 – “Screenshot of the reality I see” 3C

Source: Arpinati (2023).

This dichotomy underscores a fundamental debate on the optimal utilization of AI’s

capabilities while ensuring that ethical standards in scholarly work are not compromised. This
division points to the critical need for developing and implementing AI-driven ethical
safeguards that uphold writing integrity. It is crucial that these tools are used responsibly within
academic settings to maintain trust in educational outcomes and scholarly publications.
Through such measures, academia can navigate the challenges posed by AI, fostering an
environment where technological advancements and ethical standards coexist, aligning with the
dialogic principle.
AI tools play a role in maintaining the integrity and authenticity of written content
through various approaches and methodologies. One significant contribution is the use of
advanced models like BERT to identify AI-generated content by analyzing contextual
embeddings and revealing complex patterns indicative of AI origin, thus safeguarding the
integrity of human-authored content (Chakraborty et al., 2024). Co-audit tools have emerged
to help users double-check AI-generated content for correctness, especially in outputs like
summaries, tables, or code, guaranteeing that errors are minimized and quality is preserved
 97

(Gordon et al., 2023; Tamilselvi et al., 2023). AI-powered text-checking services like QuillBot
and HiveModeration have demonstrated efficacy in detecting AI-generated content, although
they sometimes fail to identify AI signs in original texts, highlighting the need for continuous
improvement in these tools (Lukianenko; Shastko; Korbut, 2024; Tamilselvi et al., 2023).
QuillBot, Grammarly, and Wordtune are AI-powered writing tools designed to enhance
various aspects of writing. QuillBot specializes in paraphrasing and summarizing text to help
avoid plagiarism and improve clarity. Grammarly acts as a comprehensive writing assistant,
offering corrections and suggestions for grammar, style, and tone, making it invaluable for
producing polished and professional text. Wordtune focuses on improving readability and style,
offering sentence restructuring and word choice enhancements to better engage and
communicate with readers. Together, these tools are beneficial for educators, students, content
creators, and professionals seeking to refine their writing quality and promote academic
integrity. Their integration, however, necessitates the adoption of ethical standards to prevent
misuse (Tamilselvi et al., 2023).
These approaches and methodologies integrate the recursive and dialogic principles
(Morin, 2015). The recursive principle is evident as the system uses continuous feedback loops
to refine and enhance its processes. Each iteration of AI analysis and human review feeds back
into the system, progressively improving its ability to detect and differentiate AI-generated
content. Simultaneously, the dialogic principle is demonstrated through the ongoing dialogue
between human cognition and AI processing. This symbiotic interaction is also captured in the
screenshot 5A (see Figure 34), which addresses the transformative potential and challenges
GenAI brings to educational integrity. “While some regard AI as a potentially transformative
technology, creating a more inclusive and personalized education, for others it makes it
impossible to trust coursework grades” (excerpt from the screenshot 5A).
 98

Figure 34 – “Screenshot of the reality I see” 5A

Source: Morrison (2024).

In the referenced article, researchers at Cambridge University have identified distinctive

characteristics of essays written with the assistance of ChatGPT. They observed a high
frequency of Latin-derived words, pleonasms, and a consistent use of the Oxford comma.
GenAI assists in generating informative and reflective content but is inadequate in areas of
analysis and comparison. The small-scale study highlighted how students use AI, ranging from
direct copying to using prompts for further research. Despite the advantages of quick
information access, there is a consensus that relying solely on AI could compromise academic
standards. These findings state the necessity of integrating GenAI into educational practices to
uphold ethical standards and maintain academic norms (Afifah, 2024), while also highlighting
the role of fostering multiliteracies to achieve these goals (Kalantzis; Cope, 2023).
The swift adoption of GPT-based text generators has required critical thinking and
mindful use in academic writing, underscoring the importance of distinguishing between human
and AI-generated content (Dingemanse, 2024). As LLMs continue to advance, efforts to detect
AI-generated content in academic settings have intensified, with research focusing on the
functionality and accuracy of detection tools to verify the authenticity of academic work.
Additionally, conferences and discussions on integrity, open science, and AI’s role in academia
 99

highlight the importance of these tools in maintaining the integrity of written content
(Dubinsky; Simón‐Soro, 2023).
Machine learning models like the Generative Textual Likelihood Ratio (GLTR) have
been proven to improve the detection rate of AI-generated text. The GLTR models enhance the
human detection rate of AI-generated text from 74% to 99%, which is crucial in educational
settings to prevent plagiarism and ensure the authenticity of student submissions (Prajapati et
al., 2024). Furthermore, studies evaluating the effectiveness of online detection tools such as
GPTZero, ZeroGPT, Writer ACD, and Originality reveal significant variability in their
accuracy, with some tools like Originality showing enhanced sensitivity to AI-generated text
patterns, indicating the need for more refined detection methodologies (Bellini et al., 2024).
In scientific research, a three-dimensional model consisting of transparency, integrity,
and accountability has been proposed to define the responsible use of AI, along with “AI Usage
Cards” to standardize reporting and promote ethical AI usage, ensuring the integrity of scientific
outputs (Wahle et al., 2023). Journals and publishers also leverage AI to screen submissions for
potential misconduct, such as plagiarism and data manipulation, ensuring the integrity of
published manuscripts while emphasizing the importance of human review to verify AI-
identified cases and maintain accountability (Hosseini; Resnik, 2024).
Editorial policies have been developed to guide the appropriate use of AI in authorship,
peer review, and editorial processing to prevent the erosion of scientific integrity (Leung et al.,
2023). Additionally, the gap between AI-generated and human-written scientific texts in terms
of depth and quality highlights the need for ongoing optimization of AI models to produce high-
quality content while addressing ethical and security concerns (Ma et al., 2023). Finally, the
vulnerability of existing plagiarism detection algorithms to AI-based tools underscores the
importance of developing more robust methods to detect and prevent academic dishonesty
(Xiao; Chatterjee; Gehringer, 2022).
In the creative sector, AI tools democratize the generation of high-quality creative
outputs, allowing individuals without traditional expertise to express their aspirations and
emotions. This sustains the perceived intentionality and authenticity of creative works (Moura,
2023). However, the rapid adoption of AI-generated content challenges authenticity,
necessitating the use of watermarking techniques to detect and prevent misuse. These methods,
though promising, require further refinement to be fully effective (Jiang; Zhang; Gong, 2023).
Through the lens of Complex Thinking (Morin, 2015), the holographic principle enhances this
landscape by suggesting that each piece of AI-generated content, despite being a single entity,
encapsulates comprehensive creative dynamics. This broader view reflects that individual
 100

works are not isolated creations but integral parts of the collective creative domain, influencing
and mirroring the entire artistic ecosystem.
Collectively, AI tools are being tested for their effectiveness in detecting artificially
generated information across multiple domains, with varying degrees of success, highlighting
the need for continuous improvement in detection technologies to maintain content authenticity
(Akram, 2023). This context issues a demand for parallel developments in education. The
“Writer’s Integrity” framework enhances this by monitoring the writing process rather than the
final product, capturing the distinct behavioral footprint of human authorship, validating human
intellectual work and preserving academic integrity and intellectual property rights (Aburass;
Abu Rumman, 2024).
The integration of GenAI tools such as ChatGPT into educational settings requires
practices that focus on the creation of detailed and effective prompts to facilitate meaningful
human-AI interactions (dos Santos; Boa Sorte; de Oliveira, forthcoming). These prompts enable
functional and ethical use of AI text generators and foster critical thinking skills. By serving as
a scaffolding tool, GenAI helps students overcome language barriers and cognitive challenges,
enhancing their creativity and collaboration. This approach also empowers students with the
necessary expertise to navigate adeptly in an AI-driven world, teaching them the skills needed
to critically analyze AI-generated content and use it effectively in their learning and writing
processes.
Considering these developments, the phenomenon of writing impacted by GenAI is a
complex system when viewed through the lens of the recursive, dialogic, and holographic
principles (Morin, 2015). Each piece of AI-generated content is not simply a static text but part
of a broader, dynamic system in which content is continuously created, interpreted, and
recreated, thereby mirroring the recursive nature of complex systems. Similarly, the dialogic
principle emphasizes that AI-generated texts facilitate ongoing conversations between human
and machine, each learning and adapting from the other.
To expand on this view, the holographic principle demonstrates that in complex
systems, each part reflects the entire system. This principle shows how individual interactions
with GenAI represent and actively influence overarching educational goals, such as fostering
critical thinking and ethical engagement. These interactions are more than reflections. They are
formative elements that help shape both the educational framework and cognitive development.
Recognizing writing impacted by GenAI as a complex system implies that this phenomenon
continuously evolves with technological advancements. It reveals the need for educational
 101

practices that develop our ability to engage both critically and creatively in this evolving
landscape.

3.3 Writing education is transformed by AI through personalization and global

reach

The transformation of writing education through AI is represented in the construct

WRITING EDUCATION IS TRANSFORMED BY AI THROUGH PERSONALIZATION
AND GLOBAL REACH. This theoretical construct explores how AI reshapes educational
experiences by providing personalized learning opportunities and extending the accessibility of
educational content across diverse languages and regions. The themes elaborated here illustrate
a significant evolution in writing education, emphasizing the capabilities of GenAI to cater to
individual learning needs and bridge educational gaps globally. This shift offers possibilities
for enhancing the learning experience and for democratizing access to quality education
worldwide.
AI’s ability to personalize learning experiences and extend educational content globally
represents a substantial advancement in writing education, offering new opportunities for
students and educators alike to engage with and benefit from cutting-edge technologies (Brady
et al., 2023). This global reach is supported by the increasing number of research papers on AI
in education, highlighting the growing interest and collaboration in this field (Paek; Kim, 2021).
However, the integration of AI in education is not without challenges. Issues such as
algorithmic bias, privacy concerns, and overreliance need to be addressed to ensure the
responsible use of AI technologies (Abd-Alrazaq et al., 2023).
Complex Thinking (Morin, 2015) can foster substantial insights, particularly through its
principles, in exploring the impact of GenAI on writing education. Recursively, AI systems
refine their approaches based on ongoing feedback from student performance, continuously
improving the educational process. Dialogically, the expanding body of research and
international collaboration provides diverse insights that address critical challenges such as
algorithmic bias and privacy concerns. Holographically, localized implementations of AI in
writing education reflect broader global trends toward enhanced and accessible educational
practices, highlighting the interconnected nature of these advancements. This framework
facilitates a comprehensive understanding of AI’s impact on writing education, emphasizing
the necessity for a balanced and informed approach to technology integration.
 102

In medical education, AI has shown significant promise in improving learning

outcomes, such as in the AI-assisted medical image learning system, which enhanced diagnostic
accuracy among students (Cheng et al., 2020). This indicates the broader applicability of AI in
various educational fields, including writing.
GenAI’s capacity to provide timely and personalized feedback in educational
environments, such as English as a Foreign Language (EFL) classrooms, encourages student
involvement and improves grading efficiency, highlighting its role in enriching the instruction
and acquisition of writing skills (Dong, 2023). However, the use of GenAI in writing education
should be balanced with the intrinsic value of human writing experiences, focusing on the
process rather than just the production of texts (Yoon, 2023).
The incorporation of GenAI in EFL classrooms illustrates Complex Thinking principles,
particularly the holographic and dialogic. GenAI’s role in providing personalized feedback
boosts student involvement and efficiency, aligning with the holographic principle where each
component reflects broader educational goals. Meanwhile, the dialogic principle surfaces in the
ongoing balance between the benefits of GenAI and the intrinsic value of human writing
experiences. This dynamic interaction ensures that technology enhances the educational process
without diminishing the essential human elements of creativity and personal expression in
writing.
The integration of AI in education, especially through LLMs like ChatGPT, has the
potential to transform how students learn and engage with writing tasks. AI-driven tools can
provide personalized feedback on assignments, as demonstrated by the Personalized
Multimodal Feedback Generation Network (PMFGN), which generates accurate and diverse
feedback by integrating multimodal inputs such as text, images, and audio (Liu et al., 2020).
This personalized approach is crucial in addressing individual learning needs and enhancing the
educational experience.
AI’s ability to analyze and optimize writing patterns, as exemplified by the Intelligent
Online Assessment (IOA) English analysis model, helps students improve their writing skills
through structured peer assessments and targeted feedback (Wu, 2023). These AI-powered tools
leverage advanced ML and NLP algorithms to detect nuances in language use and structural
integrity in students’ writings. This enables them to offer precise corrections and actionable
suggestions that improve students’ understanding of academic English norms. Further, by
automating the review process, these tools allow for more frequent and timely feedback, crucial
for iterative learning processes. This model enhances students’ writing proficiency and fosters
 103

a deeper engagement with the mechanics of writing, such as grammar, style, coherence, and
content as indicated in recent studies.
Additionally, AI’s role in education extends beyond personalized learning; it also offers
possibilities for facilitating the global dissemination of educational content. The advancements
in NLP and transformer language models (TLMs) enable the creation of multilingual
educational resources, making high-quality content accessible to a global audience
(Gruetzemacher; Paradice, 2022). This potential benefit is also captured in the screenshot 5B
(see Figure 35), an interview with Sal Khan, the founder and CEO of Khan Academy13. While
discussing the broader implications of AI on education, he presents Khan Academy’s
impressive reach and impact: over 160 million registered users across more than 190 countries
and availability in over 50 languages.

Figure 35 – “Screenshot of the reality I see” 5B

Source: CNBC Television (2024).

13
Khan Academy is an internet-based educational platform that is freely accessible, offering instructional videos,
practice exercises, and customized learning dashboards across various subjects spanning a variety of academic
disciplines. Through the utilization of sophisticated GenAI technologies, such as GPT-4 via its proprietary tool
known as Khanmigo, the platform customizes the educational journey for each student, enabling them to
advance at their own speed and targeting specific areas of academic challenge. Funded by donations and grants,
Khan Academy aims to provide a high-quality education regardless of their geographic location, although
accessibility might be constrained by the availability of internet connectivity and technological devices.
Available at: [Link] and [Link] .
 104

Despite the widespread reach of Khan Academy, leveraging AI and the internet to
democratize education, the digital divide presents significant challenges in achieving truly
universal access. A critical obstacle is the lack of reliable internet access for many individuals
globally, which is fundamental to utilizing online resources like Khan Academy. Additionally,
the absence of necessary technological devices such as computers, tablets, or smartphones
limits the access of many students to these educational tools.
Beyond these digital gaps, broader societal inequities like poverty, inadequate
infrastructure, and political instability severely impede educational access. Addressing these
issues demands a collaborative approach among governments, non-governmental
organizations, and other key stakeholders to forge an equitable educational landscape that
extends beyond digital solutions.
This promise for democratizing access to quality education worldwide, often
highlighted by proponents of educational technology, warrants critical examination. It raises
questions about the actual efficacy of AI in bridging educational disparities across diverse
geographic and socio-economic contexts. Such scrutiny is essential to discerning whether AI
can truly deliver on its potential to democratize access to quality education globally.
The results from the three theoretical constructs addressed the first specific objective of
this research. The key terms — revolution, integrity, and transformation — extracted from each
construct, convey the approach to the phenomenon of writing on internet-based platforms,
particularly after the launch of ChatGPT. According to Complex Thinking (Morin, 2015),
revolution, integrity, and transformation are interrelated phenomena that interact within a
dynamic system, collectively influencing and shaping the broader phenomenon of writing on
internet-based platforms, especially after the launch of ChatGPT, by driving profound shifts in
writing practices, upholding ethical standards, and continuously evolving writing processes.
 105

SECTION 4

EDUCATIONAL, PEDAGOGICAL, AND

ETHICAL ISSUES OF AI TEXT
GENERATORS14

14
Image generated using OpenAI’s DALL-E image generation tool on November 11, 2023. The prompt used for
generation was: “Generate an image that shows human and machine integrating to write.”
 106

4 EDUCATIONAL, PEDAGOGICAL, AND ETHICAL ISSUES OF AI TEXT

GENERATORS

“Writing words by hand is a technology

that’s just too slow for our times”
(Dibbell, 2010, p. 174).

This section presents the 25 revealed themes that were fundamental to the development
of the constructs discussed herein (see Figure 36). In this analytical path, the “screenshots of
the reality I see” interpreted through Complex Thinking principles (Morin, 2015) exemplify the
interconnectedness of the phenomena studied in this research: writing, education, and GenAI
technologies.
 107

Figure 36 – Theoretical Constructs: specific objective 2

Source: created by the author (2023).

 108

I initiate the presentation and analysis of the research findings, addressing the second
specific objective of the study. This objective concerns the phenomenon of writing, recognized
as a life experience (Ravagnoli, 2017), being impacted by the phenomenon of AI text
generators, particularly ChatGPT. This influence spans three critical areas: educational,
pedagogical, and ethical issues. Educational concerns focus on the broad impact of AI text
generators on the educational landscape, including aspects such as assessment. Pedagogical
issues relate specifically to strategies, methods or theories of teaching writing. Ethical
considerations explore the moral implications, questioning what is right or wrong in the use of
such technology in writing practices. The data supporting this analysis, “screenshots of the
reality I see”, is derived from the documents found on internet-based platforms, phenomena
that capture writing as a social practice. This understanding is underpinned by the principles of
Complex Thinking: recursive, holographic, and dialogic (Morin, 2015).
Recognizing writing impacted by GenAI as a complex system is acknowledging it as a
phenomenon whose dynamic nature continuously evolves with technological advancements.
The recursive principle (Morin, 2015) reveals how educational impacts of AI on writing are not
isolated events but rather cyclical processes that affect and are affected by the educational
system at large. For example, the integration of AI in assessment practices both shapes and is
shaped by evolving educational standards and learning outcomes. In terms of pedagogy, the
holographic principle (Morin, 2015) applies where each part of the teaching methodology
reflects the whole pedagogical framework. This is evident in how AI text generators like
ChatGPT can be used to complement traditional teaching methods, thereby enhancing the
holistic approach to writing instruction. This integration demonstrates how individual
pedagogical techniques influenced by GenAI mirror the overall objectives and methodologies
of educational institutions. Ethical issues are examined through the dialogic principle (Morin,
2015), emphasizing the importance of ongoing dialogue concerning the moral dimensions of
using GenAI in writing. This dialogue involves multiple stakeholders — including educators,
students, and policy makers — engaging in discussions about what is deemed right or wrong in
the use of GenAI technologies. This principle encourages a broad and inclusive conversation
that acknowledges diverse perspectives and the evolving nature of ethical standards in a
digitally enhanced educational landscape.
This section is structured into five subsections, each dedicated to a distinct theoretical
construct, elaborated through the movements of Analytical Entwining integrated with GenAI
and in collaboration with ChatGPT. The exploration of each construct commences with a
delineation of its essential themes. The following analysis includes references to scholars who
 109

have contributed to the debate on these constructs, incorporating excerpts from literature and
research data to enrich the discussion. This approach is uniformly applied across all the five
constructs.

4.1 Writing enhances educational practices with AI tools

The theoretical construct WRITING ENHANCES EDUCATIONAL PRACTICES

WITH AI TOOLS asserts that the integration of AI technologies into writing instruction
improves educational outcomes. This is supported by themes including the enhancement of
essay quality through AI-driven insights, the development of writing skills via structured AI-
enhanced approaches, and increased engagement through interactive AI learning tools.
Additionally, AI tools offer personalized tutoring experiences and promote critical thinking and
editing skills.
Written language is a technology learned and produced in social contexts, enabling the
establishment of social bonds, construction of shared meanings, changes in the social presence
of the one who writes and the realization of social actions. Writing is a social technology
developed in order to serve communication between people, and its social value motivates both
the difficult learning of material and symbolic technologies related to writing and the arduous
psychological effort devoted to the act of writing (Bazerman, 2016).
In exploring the evolution of writing, it is evident that technological advancements have
significantly reshaped cultural practices associated with writing, particularly within educational
contexts. The proliferation of social media has popularized and fundamentally transformed
writing into one of the most dynamic cultural practices today. In English language classrooms,
this transformation is characterized by the integration of multimodal compositions and the
critical engagement with remixed texts. This shift encourages educators to foster a curriculum
that transcends traditional writing modalities, advocating for the inclusion of multiliteracies that
mirror the nuanced demands of contemporary communication (Boa Sorte, 2018).
As we transition into an era dominated by AI literacy, it is crucial for educators and
students to grasp the unique, non-human aspects of how GenAI functions, and to understand its
challenges and limitations. This shift necessitates a new literacy agenda that is tailored to the
emerging realities of the GenAI landscape. This new literacy framework should encompass a
broader understanding of written text, to include the full range of Unicode characters used in
contemporary texts, such as emojis, icons, and ideographs, linking writing with mathematics
and coding. While literacy educators may not require deep expertise in these specialized fields,
 110

it is imperative that mathematics and computer science educators integrate literacy education
into their things. This holistic approach would make every educator a literacy teacher in the
GenAI era. Moreover, the advancement of multimodal literacy, which merges traditional text
with other media, acknowledges the historical multimodality of literacy and adapts to the
nuanced transitions introduced by GenAI, which can generate diverse content forms from
textual inputs (Kalantzis; Cope, 2024). For instance, the screenshot 1E (see Figure 37) provides
strategies on utilizing ChatGPT to enhance essay writing skills, exemplifying these expanded
capabilities in practice.

Figure 37 – “Screenshot of the reality I see” 1E

Source: Dr Alex Young (2023).

The screenshot 1E highlights a common challenge many students face in essay writing,
noting the lack of formal instruction with the statement: “No one really teaches us how to write
essays and getting a 1st can seem daunting”. It introduced a technique that employs a structured
approach, significantly aided by GenAI technologies like ChatGPT, to demystify and improve
essay writing. This method aims to clarify the process and increase both the efficiency and
effectiveness of writing, which is crucial for students aiming for high grades. The integration
of GenAI in this technique exemplifies the transformative impact of technology on educational
practices, particularly in refining and enhancing more traditional academic skills.
 111

AI tools have substantially enhanced writing skills and educational outcomes in

academic settings by providing multifaceted support throughout the writing process. These
tools assist in planning, drafting, editing, and revising, thereby addressing common challenges
faced by writers of all proficiency levels. For instance, AI tools like ChatGPT can generate
prompts and provide corrective feedback, improving learners’ writing skills by suggesting
enhancements in vocabulary usage, syntax, and discourse structure (Duncanson, 2024).
Through the lens of Complex Thinking (Morin, 2015), recognizing the coexistence of
dualities—human and machine—in the phenomenon of writing impacted by GenAI is crucial.
This awareness helps students understand what it means when we say ChatGPT can generate
human-like texts. The dialogic principle (Morin, 2015) suggests that dualities and
contradictions, while seemingly at odds, are not only inevitable but essential for a nuanced
understanding of phenomena. This concept rejects simplified views of reality, which often
perceives things in “black and white”, promoting instead a more complex understanding where
contradictions coexist and inform each other. Embedded in this principle is the idea that
opposition serves as a driving force in human thought and broader conceptual debates.
Engaging with opposing concepts as a form of participatory problem-solving underpins the very
nature of dialogic interaction (Bazerman, 2016).
Considering the breakthrough in deep learning, ChatGPT’s text generation exemplifies
this coexistence of human-like creativity and mechanical computation. Texts are generated by
selecting the next word based on probabilities, demonstrating a unique integration of human
cognitive patterns with machine efficiency (Wolfram, 2023). ChatGPT operates with whole
words, drawing from a corpus of approximately 40,000 common English words, and starts
generating sentences where each word is independently picked at random, mirroring the
probability of its occurrence in a vast corpus sourced from millions of books. As an LLM,
ChatGPT estimates sequence probabilities, using a model equipped with sufficient parameters
to compute next-word probabilities effectively enough to produce reasonable essay-length
pieces of text. This model functions through neural networks—simple idealizations of how
human brains operate, arranged in layers of connected “neurons”. These artificial neural
networks capture a human-like way of processing information (Wolfram, 2023), reflecting a
complex interaction of simplicity and complexity. Furthermore, the structure and functioning
of these neural networks embody the holographic principle (Morin, 2015), where each
 112

component performs a specific function and simultaneously contributes to the holistic

capability of the model, thus mirroring the overall cognitive process of language generation.
In educational contexts, AI tools have been leveraged to develop course materials and
assist students in interacting with statistical and data science concepts responsibly, drawing
parallels to the initial skepticism and eventual acceptance of calculators in classrooms (Ellis;
Slade, 2023). AI-generated writing tools like ChatGPT present challenges that require
pedagogical frameworks to teach students how to effectively partner with AI, ensuring that they
can corroborate and incorporate AI-generated content responsibly (Tseng; Warschauer, 2023).
The screenshot 2E (see Figure 38) discusses an educational strategy in which a university
professor has students correct errors in essays written by ChatGPT.

Figure 38 – “Screenshot of the reality I see” 2E

Source: @filipedeeschamps (2023).

The professor’s method involves having students identify and correct mistakes in texts
generated by ChatGPT. As highlighted in the passage, “The publication about the case on
Reddit attracted many positive comments, with users congratulating the teacher for addressing
technology in a way that demonstrates the tool’s limitations, as well as teaching students to
better understand writing” (my translation). This strategy exposes the limitations of GenAI in
writing and actively engages students in the learning process, enhancing their understanding of
 113

writing mechanics and fostering critical thinking skills. The enthusiastic reception on Reddit
underscores the potential of integrating GenAI tools in education to both challenge and educate
students more effectively. Similarly, the application of these tools reflects the principles of
complex systems (Morin, 2015), where interventions, such as teaching students to critically
analyze AI-generated texts, initiate feedback loops that enhance learning outcomes and adapt
pedagogical approaches.
The recursive nature of complex systems (Morin, 2015) makes them less predictable
and controllable, as changes in one part of the system can have unforeseen effects elsewhere
due to the continuous feedback loops. Unlike linear causality, which flows in one direction—
from cause to effect — these systems exhibit circular causality, in which causes and effects
continuously influence each other in a loop. Systems characterized by recursive thinking often
involve feedback loops. Positive feedback loops can lead to exponential growth or escalation,
while negative feedback loops tend to stabilize systems.
Incorporating the concept of the recursive principle (Morin, 2015) in education
encourages students to think beyond linear narratives and understand the complexity and
dynamism of the world around them. This principle is key in understanding complex systems
in which simple cause-effect relationships are insufficient. It emphasizes the interconnectedness
and interdependence of elements within a system, representing a shift from linear,
compartmentalized knowledge to a more integrated, holistic understanding. Recognizing this
principle allows for a deeper comprehension of the dynamic and evolving nature of complex
systems. The screenshot 3E (see Figure 39) illustrates how 1st and 5th graders are using the
SchoolAI tool to learn how to formulate their own questions, demonstrating a unique
application of AI in educational settings.
 114

Figure 39 – “Screenshot of the reality I see” 3E

Source: @hollyclarkedu (2024).

The initiative, highlighted by educators in San Diego (USA), leverages SchoolAI to

engage students more deeply in their learning processes. The tool enables students to ask their
own questions, fostering a sense of ownership and curiosity about the material. As noted in the
educator’s enthusiastic feedback, “The tool —SchoolAI in this case—had students totally
invested in content because they could ask their own questions!” This approach encourages
critical thinking and enhances questioning skills (Freire, 1985; Freire; Faundez, 1992), and also
ensures that screen time is spent on meaningful, interactive learning. It represents a step forward
in utilizing technology to enrich education, promoting responsible and collaborative use of AI
among young learners.
AI tools have the potential to substantially improve writing affordances and educational
outcomes in academic settings by providing multifaceted support to both students and
educators. LLMs, like ChatGPT, have played a crucial role in revolutionizing the process of
information retrieval and comprehension, closely related to the influence exerted by Google.
These tools enable users to pose questions and receive thorough explanations, thus facilitating
understanding over intricate concepts (Davison et al., 2023).
Regarding these transformative shifts in multiliteracies, in which traditional and digital
modalities increasingly converge, it is important to consider the evolving perceptions of more
traditional practices. Reflecting on the epigraph at the beginning of this section (Dibbell, 2010),
 115

the notion that writing by hand is seen as too slow for our times highlights a significant cultural
shift. This change remains particularly profound as we continue to navigate the transition from
paper to digital screens, a transformation propelled by the rise of computers and the internet.
However, this transition is not experienced uniformly; many individuals and communities still
lack access to computers and the internet, resulting in a growing digital divide. Advanced text-
generating AI technologies are setting new standards for speed and efficiency, further widening
this gap. This disparity is exacerbated by the unequal implementation of educational public
policies across different countries and schools.
The transformative shifts in writing practices, as detailed in the preceding discussions,
can be interpreted through the lens of Complex Thinking (Morin, 2015), aiding in
understanding the reorganization processes the phenomenon of writing is undergoing. The
holographic principle is evident as every part of the practices of multiliteracies (Kalantzis;
Cope; Pinheiro, 2020; Kalantzis; Cope, 2023) reflects the overall cultural shift towards digital
integration; small changes in classroom approaches mirror larger societal shifts towards
multimodal communication. The recursive principle is observed in the iterative process of
adapting literacy curricula to include traditional texts, digital content, and AI-generated content,
suggesting a continual revisiting and refining of educational practices. The dialogic aspect is
prominent in the engagement with multiliteracies, where texts are not merely consumed but
interacted with, creating a dynamic exchange between students, educators, and digital content.
Finally, the notion of a system is crucial, as literacy no longer stands alone but is interconnected
with various fields such as mathematics, coding, and AI, forming an integrated network of
knowledge that transcends traditional boundaries. These principles guide the understanding that
writing practices are evolving into more complex, integrated systems that reflect the intricate
realities of the contemporary world.

4.2 Writing processes supported by AI’s automation and personalization

The theoretical construct WRITING PROCESSES SUPPORTED BY AI’S

AUTOMATION AND PERSONALIZATION embodies a multifaceted approach to
understanding how AI is reshaping writing practices. Central to this construct is the premise
that AI’s capabilities can significantly enhance writer’s efficiency through the automation of
assessments and personalized feedback, streamlining the writing process and potentially
improving the quality of outputs. Furthermore, the use of AI tools in writing facilitates the
generation of diverse and creative content, catering to a broad spectrum of writing requirements
 116

and styles. However, this technological integration is not devoid of challenges. The detectability
of AI-generated content poses significant concerns for writing assessments, complicating the
ability to discern human from machine-generated text. Additionally, the use of AI in completing
writing assignments introduces ethical dilemmas, particularly concerning originality and
authorship. This shift suggests a transformation in the role of the writer, predominantly towards
revising and editing AI-generated drafts, thereby redefining traditional writing dynamics and
responsibilities.
From the perspective of Complex Thinking (Morin, 2015), this theoretical construct
highlights the recursive, dialogic, and systemic nature of incorporating AI into writing. This
integration demonstrates the recursive principle by showcasing how AI tools and human input
continually influence and reshape each other. Writers adjust their strategies based on AI
feedback, while AI systems evolve through interactions with human-generated content, creating
a continuous loop of mutual enhancement and adaptation.
The dialogic principle (Morin, 2015) emerges in the exchange between technology and
traditional writing practices, in which AI capabilities challenge and converse with established
norms of writing and authorship. This dialogue extends to ethical considerations, as the
community debates the implications of AI-generated content, reflecting on what constitutes
authenticity and intellectual ownership in the “post-digital” age (Cascone, 2000; Cramer, 2015).
Furthermore, examining this construct through a systemic lens (Morin, 2015) reveals
the interrelated nature of different elements such as technology, ethics, education, and societal
norms. The incorporation of GenAI into writing is not an isolated phenomenon but part of a
broader system that includes educational policies, technological advancements, and cultural
values. Each element within this system affects and is affected by the others, indicating that
changes in one aspect, like GenAI’s role in writing, could have ripple effects throughout the
entire educational and social landscape. The screenshot 1D (see Figure 40) provides a visual
representation of this discussion, highlighting the multifaceted roles of ChatGPT as a disruptor
across various industries, including education.
 117

Figure 40 – “Screenshot of the reality I see” 1D

Source: Hard Fork (2022).

In the screenshot, ChatGPT exemplifies practical applications of GenAI in the

educational sector. Serving as a personalized tutoring agent, ChatGPT adapts to individual
student needs, enhancing the learning experience through customized support. This LLM
facilitates one-on-one tutoring tailored to the pace and preferences of students and automates
assessments and feedback, significantly reducing the workload of educators. The incorporation
of ChatGPT into educational practices promises to refine teaching methodologies and augment
student engagement through personalized interaction and efficient evaluation processes. This
application of GenAI in education illustrates the dialogic and recursive principles (Morin,
2015), as it reflects the ongoing interaction between technological innovation and conventional
educational methodologies.
Similarly, the use of automation in AI-driven writing tools streamlines the writing
process, echoing the efficiencies observed with ChatGPT in broader educational contexts. By
reducing time and effort for content generation and editing, these tools offer significant benefits
in professional and academic settings in which time constraints are critical (Borg, 2021;
Davison et al., 2023). This reduction in workload is crucial, particularly in the context of writing
education in which developing cognitive skills such as defining goals, solving problems, and
 118

managing memory resources necessitates intensive practice and personalized feedback. The
demand on teachers for preparing instructional materials, and managing the labor-intensive
tasks of reading, editing, and providing feedback on student writings is significantly alleviated
(Allen; Jacovina; McNamara, 2016), enabling a more focused approach to nurturing essential
writing skills.
In accordance with the National Commission on Writing (2004), time for this process is
rare as teachers are faced with large classes further reducing time for instruction and planning.
This context justified the initial development of computer systems for summative assessment
and instruction of texts, including more recent studies that seek to develop personalized
guidelines for formative assessment based on individual analysis of performance and behaviors.
The algorithms of these so-called W-Pal systems make use of a number of different techniques
that provide personalized feedback identifying everything from more basic linguistic aspects
such as word frequency, length of each sentence to features such as strategies for rhetoric, and
semantic cohesion, elaboration, organization, and lexical sophistication (McNamara et al.,
2014). However, these systems fail to provide feedback to subjective and more significant
aspects, such as students’ creativity and development of ideas and arguments.
Building on these foundational AI systems, newer technologies, including AI-based
Automated Writing Evaluation (AWE) systems like Grammarly, enhance writing accuracy and
fluency by offering real-time feedback on grammatical errors and stylistic choices (Yan, 2023).
The integration of AI tools in writing instruction also extends to the assessment phase, in which
tools like ChatGPT can automatically grade written work, provide specific comments, and
evaluate various aspects of the text, including content richness, clarity, and organization
(Barrot, 2023). Moreover, AI tools like QuillBot are designed to bolster sentence structure and
vocabulary proficiency, making them valuable for multilingual learners (Duncanson, 2024).
As AI tools become increasingly embedded in educational practices, the necessity to
critically reflect on their uses is emphasized. This critical-reflective posture is advocated to
comprehend the possible changes in the teaching-learning context and teaching practices
brought about by algorithms designed to personalize learning, addressing each student
individually (Kaufman; Santaella, 2020; Santos; Arruda, 2019). The importance of this critical
examination is visually represented in the screenshot 4D (see Figure 41), which captures the
reactions to Governor Tarcísio de Freitas’s proposal to use ChatGPT for creating digital lessons
in São Paulo’s public schools.
 119

Figure 41 – “Screenshot of the reality I see” 4D

Source: @[Link] (2024).

The approach involves using ChatGPT to develop educational content, a strategy that
has stirred considerable debate about replacing some functions traditionally performed by
teachers with GenAI. In this screenshot addressing this issue, concerns are voiced about the
proposal being controversial and worrying: “At the very least, the proposal is controversial. In
reality, it is very worrying to see the use of AI grow in this way” (my translation). This statement
highlights the tension between technological advancement and the preservation of human-
centered teaching practices. The discussion underscores the broader implications of AI’s role
in education, emphasizing the need for a careful balance between leveraging AI for efficiency
and maintaining essential human interactions that are critical for educational and social
development.
The transformative potential of AI tools in educational practices extends beyond
ChatGPT, encompassing their applications in lesson planning, content generation, and task
creation. A conducted pedagogical experiment highlights the benefits and challenges, revealing
significant advantages and drawbacks. Critically, this research assesses teachers’ competence
in utilizing these tools, focusing on awareness, adaptability, and confidence. The findings
indicate the need for ongoing development of digital competence among teachers and a
conscious understanding of AI’s risks and limitations (Kyrpa et al., 2024). The necessity for
 120

further investigation into the educational potential of AI tools and the need to equip teachers
with effective integration skills is stressed (Bazerman, 2023; Copes; Kalantziz, 2024; Kyrpa et
al., 2024).
Following this exploration on the impact of AI tools in education, the Complex Thinking
principles (Morin, 2015) illuminate how these tools can be integrated and assessed effectively.
The holographic principle is mirrored in how individual components, such as teacher
competence, reflect broader educational impacts, suggesting that the effectiveness of AI tools
is closely interconnected with educators’ holistic understanding and implementation. The
recursive nature of this approach is underscored as the study emphasizes the need for ongoing
development in educators’ digital competencies, fostering a continuous improvement cycle
within educational practices. The dialogic principle highlights the importance of sustained
dialogue among technology developers, educators, and students to ensure that AI deployments
meet the diverse needs of the educational community and enhance learning outcomes. Lastly,
the notion of system underscores the interconnectedness of all elements within educational
ecosystems, advocating for a systemic approach to the integration of AI tools that considers all
interactions between technology, pedagogical strategies, and curriculum development. These
principles advocate for a multifaceted and dynamic approach to integrating AI in education
(Kyrpa et al., 2024).

4.3 Writing quality and authenticity in the age of AI

The theoretical construct WRITING QUALITY AND AUTHENTICITY IN THE AGE

OF AI examines the dual aspects of enhancement and challenge presented by the incorporation
of GenAI in writing processes. This construct highlights the concern that AI-generated content,
while innovative, can sometimes lack accuracy and contain biases, potentially undermining the
quality and trustworthiness of the output. It also addresses significant concerns regarding the
authenticity and originality of content, crucial in maintaining the integrity of written works in
academic and professional fields. Despite these obstacles, this construct acknowledges the
advantages of GenAI, including its capacity to foster creativity and improve problem-solving
skills through collaborative efforts with human authors. Furthermore, it underscores the
necessity of writing responsibly with GenAI by understanding its limitations and potentials,
suggesting that a nuanced approach can lead to significant improvements in content quality.
This includes leveraging AI’s advanced feedback mechanisms, which can refine content
precision and effectiveness, ultimately enriching the writing process. This construct advocates
 121

for a balanced view that embraces the technological advancements GenAI brings to writing
while critically addressing its pitfalls.
Building on this construct through the lens of Complex Thinking principles (Morin,
2015) reveals the inherently complex nature of writing systems. By applying the dialogic
principle, we recognize that writing involves constant interactions between human creativity
and GenAI capabilities. These elements should be integrated to enhance the depth and breadth
of written content. The recursive principle further illuminates how the inputs and outputs of AI
tools in writing are interdependent. Each cycle of feedback and revision potentially elevates the
quality of the final product, emphasizing the recursive nature of learning and adaptation in
writing.
The holographic principle suggests that every part of the writing process reflects the
whole system, indicating that changes in any component — be it AI-driven grammar correction
or human stylistic input — resonate throughout the entire text. Through an examination of the
notion of the system, we understand that the writing is an ecosystem, that is, it is shaped by the
confluence of multiple factors including technological tools, authorial purpose, and reader
interpretation, all of which combine to form a dynamic and evolving narrative space. Hence,
embracing these principles allows us to perceive and engage with writing not just as a linear act
of producing text, but as a dynamic, interactive system that is continually shaped and reshaped
by various influences.
GenAI technologies, such as grammar checkers and style editors, assist writers in
producing error-free and stylistically consistent texts. These tools can identify grammatical
errors, suggest vocabulary improvements, and even offer stylistic adjustments, thereby
enhancing the overall quality of writing (Davison et al., 2023). Moreover, AI-driven platforms
can analyze large volumes of text to provide insights into writing patterns and trends, which
can be particularly beneficial for academic and professional writers seeking to refine their work
(Misra; Chandwar, 2023).
In terms of creativity, AI has expanded the horizons for writers by generating content
ideas and even drafting initial versions of texts. This capability enables writers to explore
diverse perspectives and styles, fostering more innovative and engaging content. GenAI, in
particular, can simulate a variety of writing styles and genres, providing a broader palette of
creative options for authors (Stokel-Walker, 2023). The screenshot 4F (see Figure 42) illustrates
this point through a discussion on the use of AI in creative collaborations, particularly in the
context of co-authoring books. This visualization captures the dynamic interplay between
human creativity and GenAI capabilities in producing literary works.
 122

Figure 42 – “Screenshot of the reality I see” 4F

Source: @profpatriciamedeiros (2024).

This partnership exemplifies a significant evolution in the integration of GenAI in

literary creation. Highlighting the pivotal question, “How can this collaboration between human
and machine influence the quality and originality of literary works?” (my translation) the
conversation addresses the potential impacts of AI text generators on the quality and originality
of literary works. It challenges us to consider the ethical dimensions of GenAI’s role in the
creative process, questioning whether it can enhance literary output without diluting the
intrinsic human qualities of emotion and personal insight that define compelling literature. This
development opens a broader debate on the future of authorship and the authenticity of AI-
assisted creative works.
Reflecting through the lens of Complex Thinking principles (Morin, 2015), this scenario
underscores the dialogic and holographic nature of writing as a system where both human and
GenAI technologies contribute to a complex creative process. Such interactions highlight the
necessity of navigating these partnerships with a nuanced understanding of both the possibilities
and the limitations inherent in this collaboration.
However, the use of GenAI in writing also raises questions about the authenticity of AI-
generated content and the role of human creativity in writing. Beyond enhancing creative
 123

capacities, AI significantly increases accessibility, making writing tools available to a broader

audience, including non-native speakers and individuals with disabilities. Through the
provision of instantaneous feedback and recommendations, AI mechanisms have the potential
to assist users in refining their writing proficiencies and crafting superior quality texts,
irrespective of their initial skill level (Májovský et al., 2023).
The democratization of writing tools has the potential to foster a more inclusive
environment in which a variety of voices are acknowledged and appreciated. Despite these
benefits, the integration of AI in writing poses challenges related to ethical considerations. The
employment of content generated by AI can blur the distinctions between human authorship
and machine output, raising concerns about the authenticity of the written text. Readers may
question whether the content reflects genuine human insight or is merely a product of
algorithmic processes (Oermann; Kondziolka, 2023).
The reliance on GenAI tools could result in a standardization of writing styles, with
writers possibly adhering to the recommendations offered by such tools, potentially inhibiting
personal expression and originality (Dupps, 2023). Additionally, ethical considerations come
to light in connection with the use of AI in writing, encompassing issues like data
confidentiality, protection of intellectual property rights, and the risk of AI-generated
misinformation, all of which require resolution to guarantee the suitable utilization of these
technologies (Leopold et al., 2023).
Furthermore, the potential for AI to perpetuate biases present in training data is a
significant concern, as it can lead to biased writing outputs that reinforce existing stereotypes
and inequalities (Peng; May, 2023). Through the lens of Complex Thinking (Morin, 2015),
these dynamics reflect the recursive and holographic principles, in which the interaction
between technology and societal norms creates a feedback loop that can either perpetuate or
challenge existing disparities.
GenAI has a profound impact on contemporary writing quality, offering numerous
benefits in terms of efficiency, creativity, and accessibility. Nevertheless, it also brings forth
challenges concerning authenticity, ethical dilemmas, and the susceptibility to bias. In the
ongoing evolution of GenAI, it is imperative for authors, academics, and decision-makers to
adeptly maneuver through these intricacies to leverage the complete capabilities of AI while
ensuring the preservation of the authenticity and inclusivity of human communication in written
form. By embracing the dialogic, recursive, and holographic principles of Complex Thinking
(Morin, 2015), stakeholders can better understand and manage the intricate relationships and
impacts of AI integration in the literary field.
 124

4.4 Writing and educational equity with AI tools

The theoretical construct of WRITING AND EDUCATIONAL EQUITY WITH AI

TOOLS explores the potential of AI to foster or hinder equity in educational settings through
its integration into writing practices. Central to this construct is the imperative to ensure that AI
writing tools are universally accessible, thereby preventing exacerbations of existing
educational inequalities. In academic contexts, the deployment of AI can be optimized through
workshops and practical demonstrations, which educate and empower students and educators
to effectively utilize these technologies. AI’s capabilities also extend to streamlining the
creation of teaching materials, significantly reducing the workload on educators and enabling
more focused, personalized instruction. Collaborative writing projects, especially on a large
scale, can benefit immensely from AI’s organizational and generative capacities, facilitating
efficient project management and execution. Moreover, while AI can serve as a catalyst for
generating new ideas, maintaining academic integrity is fundamental, requiring a balanced
approach to the use of AI in order to uphold the values of originality and critical thinking in
writing.
Through the lens of Complex Thinking (Morin, 2015), the integration of GenAI into
writing practices is perceived not simply as a technological enhancement but as a complex
transformation interacting with the systemic, personal, and pedagogical dimensions of
education. By invoking the holographic principle, we see that each part of the educational
system reflects and contains the whole, suggesting that AI tools should be designed and
implemented to mirror the entire educational ethos, including considerations of equity and
accessibility. Additionally, the recursive principle underlines the ongoing feedback loops
between the use of AI tools and educational outcomes, where each implementation can refine
both the tools themselves and their impact on education. This perspective allows us to
understand the deployment of AI in educational settings as a dynamic and evolving system,
necessitating a nuanced consideration of its widespread effects on educational equity.
Reflecting on the ambitious potential of GenAI, critical assessments address the
historical and ongoing challenges of using technology to foster educational equity (Kalantzis;
Cope, 2024). Despite the widespread integration of computers in learning environments since
the early 21st century, there has been little significant impact on the deep-rooted issues of
educational and social inequality. Literacy, a crucial indicator and potential perpetuator of these
disparities, remains an enduring challenge. This context raises crucial questions: “Can
Generative AI help change the game? Can it help calibrate learning to address the great
 125

differences between students across many dimensions? Can inexpensive, one-to-one, AI-
supported literacy teaching close the gap?” (Kalantzis; Cope, 2024, p. 27).
The integration of AI writing tools in education presents both opportunities and
challenges in achieving educational equity. AI writing tools possess the capacity to promote
equal access to exceptional writing aid, yet they pose a potential threat of amplifying current
disparities if deployed without careful consideration. These tools have the capability to enrich
education through individualized guidance and assistance, a feature especially advantageous
for students lacking availability to conventional tutoring services. For instance, AI can offer
immediate, tailored feedback on writing assignments, helping students improve their skills at
their own pace (Pan, 2024).
This capability is especially valuable in large classrooms where individual attention
from instructors is limited. However, the effectiveness of AI writing tools in promoting
educational equity depends significantly on their accessibility and the context in which they are
used. Research indicates that disparities in access to technology can hinder the equitable
distribution of AI benefits. Students from underprivileged backgrounds may lack the necessary
devices or internet connectivity to fully utilize these tools, thereby widening the educational
gap (Stornaiuolo et al., 2023; Viberg et al., 2024). The screenshot 5F (see Figure 43) highlights
the integration of AI in education, focusing on its potential benefits and the ethical
considerations it necessitates.
 126

Figure 43 – “Screenshot of the reality I see” 5F

Source: @worldeconomicforum (2024).

In this screenshot, a leader in the field of education emphasizes the potential risk of AI
exacerbating educational disparities, highlighting the necessity of ensuring fair distribution of
AI’s benefits. This emphasizes the importance of utilizing AI in a way that caters to all students,
preventing the widening of current gaps. The promotion of fairness and inclusivity in
educational environments through meticulous planning and policy development is crucial to
guarantee that all students can access the advantages offered by AI. This concern aligns with
the holographic principle (Morin, 2015), as it reflects the need for AI tools to embody the values
of the entire educational system, ensuring that every part contributes to overall equity.
Additionally, the dialogic principle (Morin, 2015) is evident here, as it stresses the importance
of dialogue between technology and pedagogy to create an inclusive and fair educational
landscape.
Moreover, the design of AI systems often reflects the biases of their creators, which can
lead to biased feedback that disadvantages certain groups of students (Selim, 2024).
Additionally, the incorporation of AI writing aids in academic environments necessitates a
meticulous examination of instructional approaches. Educators should undergo training to
 127

adeptly integrate these tools into their pedagogical techniques, with the aim of ensuring that
they enhance rather than supplant conventional instructional approaches (Xu, 2024). This
integration holds significant importance in maintaining a balanced approach that leverages the
strengths of both AI and human instruction.
Another critical aspect is the potential for AI tools to support diverse learning needs. AI
can be programmed to recognize and adapt to different learning styles, offering customized
support that can help bridge the gap for students with varying abilities and backgrounds
(Alordiah, 2023). However, this requires ongoing research and development to ensure that AI
systems are inclusive and sensitive to the diverse needs of all learners. The ethical implications
of AI in education also warrant attention. A necessity exists for clear and open policies
concerning data privacy and the utilization of student data within artificial intelligence (AI)
systems. Ensuring that students and educators understand how these tools work and the data
they collect is essential for building trust and promoting equitable use (Roscoe et al., 2022;
Roe; Renandya; Jacobs, 2023).
While AI writing tools hold promise for enhancing educational equity, their successful
implementation requires addressing issues of access, bias, and integration into existing
educational frameworks. Policymakers, educators, and developers should engage in
collaborative efforts to develop inclusive, accessible, and efficient AI solutions that cater to the
needs of all students, especially those belonging to marginalized communities. By doing so, AI
can become a powerful ally in the quest for educational equity, providing all students with the
opportunity to succeed in their academic pursuits (Costa; Mfolo; Ntsobi, 2024; Gustilo; Ong;
Lapinid, 2024).
For GenAI to make a substantial difference, it would require innovative pedagogical
strategies and transformative changes in classroom dynamics. A scenario analysis explores both
the risk of GenAI exacerbating existing inequalities and its potential to alleviate long-standing
social divisions ingrained through traditional educational practices. Thus, a critical,
programmatic question arises: What might an agenda of educational justice look like in a time
of AI? This inquiry compels us to critically consider whether AI can be effectively utilized to
promote educational equity or if it might inadvertently solidify the status quo (Kalantzis; Cope,
2024).
Viewing the challenges and potential of GenAI through the lens of Complex Thinking
principles (Morin, 2015) enriches our understanding of its implications for educational equity.
The holographic principle compels us to consider how AI tools could reflect and reinforce the
values of equity throughout the entire educational system, rather than serving as isolated
 128

technological fixes. Applying the recursive principle reveals the importance of iterative
feedback processes in shaping the impacts of AI, ensuring that these technologies evolve in
ways that genuinely support educational justice. Moreover, the dialogic principle emphasizes
the need for ongoing, meaningful dialogues involving various stakeholders such as students,
educators, technologists, and policymakers in order to navigate the complexities of AI
incorporation and its broader societal implications. This critical examination challenges us to
question not only the capabilities of AI but also the systemic changes required to ensure it
benefits all learners equitably, avoiding the reinforcement of existing disparities.

4.5 Writing integrates AI while preserving human interaction

The theoretical construct of WRITING INTEGRATES AI WHILE PRESERVING

HUMAN INTERACTION explores the symbiosis between AI and human involvement within
the context of writing. This construct underscores that writing processes augmented by AI can
significantly enhance educational experiences by offering precision and efficiency in drafting
and editing, thus freeing up time for more creative and analytical pursuits. Crucially, it
maintains that the integration of AI in writing education should not replace but rather
complement human mentorship and interaction, ensuring that the essential elements of personal
guidance and feedback remain at the forefront of the learning experience. Furthermore,
employing AI in writing activities can make the learning process more engaging and enjoyable,
providing a dynamic and interactive environment that caters to diverse learning styles.
However, there is a critical need to strike a balance between leveraging technological
advancements and retaining traditional writing methods to preserve the depth and authenticity
of the educational experience. In scientific research, AI’s role in assisting with drafting and
editorial tasks can substantially expedite the publication process while maintaining rigorous
standards of quality and accuracy.
Through the lens of Complex Thinking principles (Morin, 2015), this construct explores
the symbiosis between AI and human interaction within the context of writing. The dialogic
principle emphasizes the continual dialogue between AI’s capabilities and human creative
processes, highlighting how AI can enhance the writing process by handling mechanical tasks,
thereby freeing human creativity for more complex and subjective aspects of writing. The
recursive principle is evident in the feedback loops between AI-enhanced tools and human
 129

input, refining both the technology and its educational application. This interaction ensures that
AI tools evolve in ways that enrich the learning experience while adhering to educational goals.
The holographic principle also plays a crucial role, suggesting that all components of
the educational system—AI tools and human mentors—should reflect and contribute to the
overarching educational objectives. By ensuring that AI tools complement rather than replace
human interaction, the educational process retains its depth and authenticity, reflecting the
comprehensive values of the system. The notion of system underscores the interconnectedness
of technological and human components within the educational ecosystem. It advocates for a
balanced approach to integrating AI, in which technological advancements and traditional
methods coexist to enhance the educational experience, catering to diverse learning styles and
needs.
Integrating AI into writing processes in educational settings offers numerous
opportunities to enhance learning while preserving essential human interactions. AI systems
can analyze student writing to identify areas for improvement, offering suggestions on
grammar, style, and structure. This can help students develop their writing skills more
effectively than traditional methods alone (Nguyen et al., 2024a; Pan, 2024). Moreover, AI can
facilitate the drafting process by generating ideas or providing templates, which can be
particularly beneficial for students who struggle with writer’s block or organizing their thoughts
(Tu et al., 2024). The screenshot 4D (see Figure 44) exemplifies this growing trend of AI
integration in educational settings, showcasing a report on the widespread use of AI
technologies for essay writing.
 130

Figure 44 – “Screenshot of the reality I see” 4D

Source: The Guardian (2024).

The surge in AI utilization among students is underscored by a survey from the Higher
Education Policy Institute, which found that “More than half of undergraduates say they consult
artificial intelligence programmes to help with their essays, while schools are trialling its use in
the classroom”. Furthermore, it was noted that 53% of those surveyed were actively using AI
to generate content for their assignments. However, direct plagiarism appears limited, with only
5% admitting to copying and pasting unedited AI-generated text into their assessments. This
highlights a shift towards using AI as a tool for enhancing academic work rather than copying
from it, suggesting a need for guidelines to optimize its use responsibly within educational
settings.
This trend underscores the dialogic principle of Complex Thinking (Morin, 2015), in
which the interaction between AI tools and educational methods evolves dynamically, pointing
towards a need for continuous dialogue and adaptation to ensure ethical integration and
effective use within academic environments. The increasing use of AI in educational settings
exemplifies a dynamic interplay between technology and educational practices. This interaction
is inherently recursive (Morin, 2015). The adaptation of AI tools in learning environments
 131

transforms student learning outcomes and necessitates ongoing adjustments in educational

strategies and policies. This recursive process requires educators and policymakers to
continuously evaluate and refine the implementation of AI to ensure it enhances rather than
undermines educational integrity. Such a cycle promotes a holistic understanding and
responsible integration of AI, aligning technological advances with the core educational values
of equity and authenticity.
However, the integration of AI should be carefully managed to ensure that it
complements rather than replaces human interactions. One of the key benefits of human
involvement in the writing process is the ability to provide nuanced feedback that considers the
emotional and contextual aspects of student work. AI systems, while efficient, may lack the
ability to fully understand the subtleties of human expression and intention (Malik et al., 2023;
Nguyen et al., 2024b). Therefore, a hybrid approach that combines AI tools with human
oversight can be most effective. Teachers can use AI-generated feedback as a starting point for
deeper discussions with students, fostering a more interactive and engaging learning
environment (Wang, 2024).
Furthermore, AI can support collaborative writing activities by facilitating
communication and coordination among students. For example, AI tools can help manage group
projects by tracking contributions, suggesting edits, and ensuring that all voices are heard. This
can enhance the collaborative experience and ensure that students learn from each other as well
as from the AI system (Costa; Mfolo; Ntsobi, 2024; Rabbianty; Azizah; Virdyna, 2023). The
screenshot 4E (see Figure 45) showcases the transformative impact of GenAI tools in education,
emphasizing their potential to enhance learning experiences.
 132

Figure 45 – “Screenshot of the reality I see” 4E

Source: Bernard Marr et al. (2024).

The screenshot highlights how GenAI can improve the way educational content is
structured and communicated, making it more accessible and engaging for learners.
“Generative AI is great for structuring and communicating information in ways that make it
easy to understand and absorb as well as providing answers; it can be used to generate
interactive learning experiences such as chatbots that can bring learning to life, or even roleplay
as historical characters”. This approach aids in delivering information and in creating dynamic
interactions that can stimulate students’ imagination and interest.
In light of Complex thinking (Morin, 2015), the use of GenAI illustrates the dialogic
nature of contemporary educational resources, in which the interaction between technology and
traditional educational practices creates a new and dynamic form of writing. This dialogic
interaction allows GenAI tools to serve dual roles: they standardize information delivery for
better understanding while simultaneously creating immersive, interactive learning modules
like chatbots. This process conveys information and engages learners in a dialogue, enhancing
retention and interest. The recursive aspect of GenAI in educational settings is evident as these
tools learn and adapt from each interaction. As students engage with AI-driven platforms, the
tools themselves evolve, continually improving their methods of information delivery and
 133

interaction. This continuous feedback loop enriches the writing process within educational
content, making it increasingly effective over time.
The holographic principle reflects how each part of the educational content — whether
a structured lesson or an interactive role-play — contains the essence of the whole educational
objective. GenAI integrates these various elements to create a cohesive learning experience that
mirrors the complexity and interconnectedness of the subject matter being taught. GenAI’s role
in education exemplifies the systemic organization, in which various components of the
educational system — content, pedagogy, technology, and learner engagement — are
integrated. GenAI synthesizes these elements, facilitating a system that adapts to and supports
diverse learning styles and needs.
Despite these advantages, there are challenges to consider. One significant concern is
the potential for AI to perpetuate biases present in the data it is trained on. This can lead to
biased feedback or suggestions, which could negatively impact student learning. It is crucial to
develop AI systems that are transparent and regularly updated to mitigate these risks (Sysoyev;
Filatov, 2024).
Additionally, there is a need for educators to be trained in effectively integrating AI
tools into their teaching practices, ensuring that they can guide students in using these
technologies responsibly and effectively (Rahayu et al., 2023). AI can be a powerful tool in
enhancing the writing process in educational settings, offering personalized feedback,
facilitating collaboration, and supporting student engagement.
Considering Complex Thinking (Morin, 2015), this theoretical construct reveals the
multifaceted nature of integrating GenAI in educational contexts, particularly in enhancing the
writing process. It advocates for comprehensive training of educators, highlighting the need for
a systemic approach to effectively harness AI’s capabilities, such as providing personalized
feedback, facilitating collaboration, and boosting student engagement. This perspective
recognizes that the incorporation of AI tools involves more than just technical deployment; it
necessitates an adaptive educational ecosystem that can foster both technological and
pedagogical innovation. By emphasizing educator preparedness, this construct stresses the
importance of human guidance in leveraging technological advancements, thereby balancing
AI’s utility with ethical considerations and educational goals. This complex interplay between
human and machine illustrates the broader implications for pedagogy and curriculum
development, urging a thoughtful integration of technology that enhances learning while
respecting the educational guidelines.
 134

The results from the five theoretical constructs addressed the second specific objective
of this research. The key terms — enhancement, support, quality, authenticity, equity, and
integration — extracted from the constructs, address educational, pedagogical, and ethical
issues related to the impact of AI-text generators on the phenomenon of writing, particularly
after the launch of ChatGPT. Through the lens of Complex Thinking (Morin, 2015), these terms
are seen as interrelated phenomena that interact within a dynamic educational ecosystem.
Collectively, they contribute to a multifaceted understanding of how AI-text generators
influence writing practices. They promote enhanced pedagogical approaches and support
personalized learning, while upholding quality and authenticity in written outputs. Furthermore,
they advocate for educational equity by ensuring that GenAI tools are accessible and beneficial
across diverse learning environments, thereby facilitating an integrated approach that respects
both technological advancements and traditional educational values. This holistic view
emphasizes the necessity of continuously adapting educational strategies to embrace the
potentials of GenAI responsibly, ensuring that these innovations serve to augment human
capabilities and foster an inclusive and effective educational landscape.
 135

SECTION 5

FINAL REMARKS15

15
Image generated using OpenAI’s DALL-E image generation tool on November 11, 2023. The prompt used for
generation was: “Basing the next generation of images on these last ones*, generate an image that expresses
the epistemological and methodological approach, Complex Thinking (by Edgar Morin), adopted for this
investigation which focuses on education, writing, generative artificial intelligence, large language models,
and ChatGPT”. *One of the last images I am referring to in this prompt is the input used for the section 2
image.
 136

5 FINAL REMARKS

“We are not passive spectators in the history of

AI, we are its authors. This means that
the values that support our visions of a future
with AI can become self-fulfilling prophecies”
(Lee, 2019, p. 271).

As I present these final remarks in my doctoral dissertation, I acknowledge that the

scope of my research does not fully capture the complexities of the phenomena studied. In light
of Complex Thinking (Morin, 2015), writing and GenAI technologies are considered
phenomena defined as emergences (Ravagnoli, 2017). These emergences arise from the
dynamic interactions among multiple system components, producing unpredictable patterns.
In this study, the phenomenon of writing is seen as a social practice, intertwined with
societal aspects. I employed these types of documents — YouTube videos, podcasts, website
and digital newspaper articles, and social media posts — as the strategy “screenshots of the
reality I see” to illustrate the interconnectedness and impact between the phenomenon of writing
and its integration with GenAI technologies. Additionally, I incorporated GenAI into my
method of analysis, which not only facilitated the analytical process but also deepened my
engagement with the phenomena under study. To summarize my research journey, I present the
key elements that guided my methodological path (see Figure 46).
 137

Figure 46 – My research journey

Source: created by the author (2023).

 138

The integration of GenAI in writing, analyzed through the principles of Complex

Thinking (Morin, 2015), reveals a complex interplay between technology and human agency.
Both elements are in constant adaptation, influencing each other in a systemic, holographic,
recursive, and dialogic manner. This interaction ensures that the writing process is dynamic,
capable of evolving with technological advancements, and remains rooted in humanistic values.
Utilizing the eight theoretical constructs developed in this study, I achieved the general
objective of this research, which is based on the nine key words associated with these constructs
(see Figure 47).

Figure 47 – The concept of writing impacted by GenAI

Source: created by the author (2023).

In the evolving educational landscape, the integration of GenAI technologies into

writing practices represents a revolution by enhancing possibilities for accessibility and
transforming the way we learn and create. GenAI tools ensure the integrity and authenticity of
written content, providing support through personalized feedback and automated assistance,
thereby transforming traditional educational approaches. These advancements enhance the
learning experience by adapting to individual needs and promoting equity across diverse student
populations. By integrating GenAI while preserving human creativity and interaction, we
achieve a balanced and ethical approach to writing that respects the core values of education
and human expression. As we move forward, it is clear that the quality of education can be
enhanced by such innovative tools that bridge gaps and foster a more inclusive environment.
 139

While the benefits of GenAI in education are substantial, they also cast a spotlight on
the pressing issue of the digital divide. The lack of access to essential technologies, including
computers and the internet, continues to widen this gap, particularly in various countries and
schools where resources are limited. This disparity impedes the equitable distribution of
educational advancements and restricts the potential for comprehensive learning and
development. To ensure equitable access to the benefits of GenAI in education, the integration
of these technologies requires robust measures and public policies that address existing
inequalities. This approach is essential for providing every student, regardless of their
geographical or economic circumstances, with equal opportunities to benefit from these
transformative tools.
Moreover, the reliance on technology alone cannot guarantee the enhancement of
writing quality. It is important to acknowledge that educational approaches, methods, and
strategies should be grounded in core elements that prioritize long-term objectives such as
critical thinking and problem-solving skills. Frequent changes in educational methodologies
with every new technological innovation can be disruptive rather than beneficial. From the
perspective of Complex Thinking (Morin, 2015), technologies, while powerful educational
resources, are not specifically designed for learning, they emerge from and are shaped by life
experience (Ravagnoli, 2017). These tools reflect and are influenced by the intentions and
contexts of their users. Therefore, employing these technologies thoughtfully is essential to
ensuring they serve as complements to, rather than replacements for, the fundamental
educational practices that foster deep, meaningful learning.
In synthesizing these perspectives, the impact of GenAI tools on the phenomenon of
writing can be meaningfully approached through the lens of Complex Thinking. This approach
illuminates how these technologies, while transformative, are part of a broader ecosystem that
includes socio-technical dynamics and human-centric values. The integration of GenAI into
writing is not merely about automating tasks or enhancing grammatical precision; it is about
enriching the writing process by fostering accessibility and addressing individual educational
needs.
As evidenced by the challenges of the digital divide and model collapse (Shumailov et
al., 2024), these technologies also introduce complexities that should be managed with care and
foresight. The influence of GenAI on writing is deeply intertwined with its ability to respect
and enhance critical thinking and problem-solving, essential human cognitive capacities. Thus,
while GenAI tools offer significant potential to transform educational practices, their
deployment should be guided by the principles of Complex Thinking, ensuring that
 140

technological advancements harmonize with the fundamental human aspects of learning and
creativity. This balanced approach leverages the benefits of GenAI while safeguarding against
its inherent risks, thereby promoting a holistic development of writing instruction in the “post-
digital” age (Cascone, 2000; Cramer, 2015).
The concept of writing with GenAI support is emblematic of a complex system where
various components — technology, users, educational frameworks, and ethical considerations
— interact within a larger educational ecosystem. This system is not static but dynamically
responds to changes within each component, affecting how writing is taught, learned, and
practiced across different contexts. In light of the ongoing debates surrounding technologies in
education, particularly AI-text generators and, more recently, the banning of smartphones in
schools, it is essential for educators and teacher educators to embrace an active, critical, and
reflective approach to writing instruction.
Educators can guide learners in harnessing the potential of these technologies by
fostering critical GenAI-supported multiliteracies and encouraging thoughtful integration of
AI-text generators. A key strategy is the effective and ethical construction of prompts to
enhance writing instruction. Thoughtfully designed, purpose-driven prompts can support
students in exploring their creativity, critical thinking, and problem-solving skills. Moreover,
prompt crafting serves as a practical approach to teaching with GenAI and an ethical means of
promoting learner autonomy in writing. It also cultivates an awareness of how to use GenAI
platforms responsibly in educational contexts. Integrating GenAI into curricula requires
thoughtful application and a critical examination of its implications to ensure it enhances, rather
than overshadows, the core values of writing as a complex phenomenon within an
interconnected ecosystem.
Recognizing writing as part of this evolving system broadens our understanding of its
multifaceted nature and challenges traditional views of writing as a static skill or mechanical
ability. This research introduces the concept of writing as an ecosystem, in which technology
and human elements are interdependent and mutually enhancing, contributing to a more
collaborative and adaptive approach to writing instruction. This reorganization embodies a
continuous interplay between technology and tradition, ensuring that the evolution of writing
practices through GenAI enriches, rather than diminishes, the educational experience.
 141

REFERENCES

ABURASS, Sanad; ABU RUMMAN, Maha. Authenticity in authorship: the writer’s integrity
framework for verifying human-generated text. Research Square, 2024. DOI:
10.21203/[Link]-4259560/v1. Accessed on: 15 June 2024.

AFIFAH. Preserving academic integrity in the age of AI. In: THE ASIAN CONFERENCE
ON EDUCATION & INTERNATIONAL DEVELOPMENT, 2024. Official conference
proceedings. ISSN: 2189-101X. Available at: [Link]
101X.2024.20. Accessed on: 11 June 2024.

AHMAD, Norita; MURUGESAN, San; KSHETRI, Nir. Generative artificial intelligence and
the education sector. IEEE Computer Society, v. 55, n. 10, p. 72-76, 2023. Available at:
[Link] Accessed on: 19 June 2024.

AKRAM, Arslan. An empirical study of AI generated text detection tools. [Link], v.

abs/2310.01423, 26 Sep. 2023. Available at: [Link] Accessed on: 4
Aug. 2024.

ALLEN, Laura K.; JACOVINA, Matthew E.; MCNAMARA, Danielle S. Computer-based

writing instruction. In: MACARTHUR, Charles A.; GRAHAM, Steve; FITZGERALD, Jill
(orgs.). Handbook of writing research. New York, NY: The Guilford Press, 2016. p. 347-
351.

ALORDIAH, Caroline Ochuko. Appreciating the AI revolution: empowering educational

researchers through AI tools for writing research articles. Zamfara International Journal of
Humanities, v. 2, n. 01, p. 178-195, 2023. DOI: 10.36349/zamijoh.2023.v02i01.013.

ATKINSON, Paul; BARKER, Richie. AI and the social construction of creativity.

Convergence, v. 29, n. 4, 05 July 2023. DOI: 10.1177/13548565231187730.

AUERBACH, Carl; SILVERSTEIN, Louise B. Qualitative data: an introduction to coding

and analysis. New York: New York University, 2003.

BARROT, Jessie S. Using ChatGPT for second language writing: pitfalls and potentials.
Assessing Writing, v. 57, 2023, p. 100745. Available at:
[Link] Accessed on: 4 Aug. 2024.

BAZERMAN, Charles. What do sociocultural studies of writing tell us about learning to

write? In: MACARTHUR, Charles A.; GRAHAM, Steve; FITZGERALD, Jill (orgs.).
Handbook of writing research. New York, NY: The Guilford Press, 2016. p. 17-31.

BENGIO, Yoshua; DUCHARME, Réjean; VINCENT, Pascal; JAUVIN, Christian. A neural

probabilistic language model. Journal of Machine Learning Research, v. 3, p. 1137-1155,
2003. DOI: 10.1162/jmlr.2003.3.1.1137.

BENZON, William. GPT-3: Waterloo or rubicon? Here be Dragons. Cognitive and

Neuroscience, Ago. 2020. Available at:
[Link]
3_Waterloo_or_Rubicon_Here_be_Dragons. Accessed on: 6 Oct. 2020.
 142

BOA SORTE, Paulo. Internet memes: classroom perspectives in the context of digital
cultures. Educação e Formação, Fortaleza, v. 4, n. 12, p. 51-66, Sept.-Oct. 2019.

BOA SORTE, Paulo. Remixes e expressão escrita em língua inglesa. In: JORDÃO, C. M.;
MARTINEZ, J. Z.; MONTE MÓR, W. (orgs.). Letramentos em prática na formação
inicial de professores de inglês. Campinas, SP: Pontes Editores, 2018. p. 279-292.

BOA SORTE, Paulo; SANTOS, Jefferson do Carmo Andrade. Memes em aulas de língua
inglesa: explorando práticas de multiletramentos. Revista Educação em Questão, Natal, v.
58, n. 55, p. 1-19, Jan.-Mar. 2020.

BOA SORTE, Paulo; FARIAS, Mário André de Freitas; DOS SANTOS, Allessandra
Elisabeth; SANTOS, Jefferson do Carmo Andrade; DIAS, Jamile Santos dos Santos
Rodrigues. Artificial intelligence in academic writing: what is in store with the GPT3
algorithm? Revista EntreLínguas, Araraquara, v. 7, n. 00, p. e021035, 2021. e-ISSN: 2447-
3529. Available at: [Link]

BOA SORTE, Paulo; SANTOS, Jefferson do Carmo Andrade; DOS SANTOS, Allessandra
Elisabeth; SILVA, Laila Gardênia Viana. SILVA, Nayara Stefanie Mandarino. Entrelace
analítico como metodologia de pesquisa em educação no paradigma emergente. Revista
Educação em Páginas, Vitória da Conquista, v. 02, p. e13531,
2023. [Link]

BODEN, M. A. Inteligência artificial: uma brevíssima introdução. São Paulo: Editora Unesp,
2020. 250 p.

BORG, Jana Schaich. Four investment areas for ethical AI: transdisciplinary opportunities to
close the publication-to-practice gap. Big Data & Society, v. 8, n. 2, 2021. DOI:
10.1177/20539517211040197.

BRASIL. Lei nº 9.279, de 14 de maio de 1996. Regula direitos e obrigações relativos à

propriedade intelectual. Brasília, DF: Presidência da República, 1996. Available at:
[Link]

BRASIL. Lei nº 9.610, de 19 de fevereiro de 1998. Altera, atualiza e consolida a legislação

sobre direitos autorais e dá outras providências. Brasília, DF: Presidência da República, 1998.
Available at: [Link]

BREMMER, Ian; STELTER, Brian; HEMMER, Nicole. Will AI influence the 2024 US
election? GZERO World, 2023. Video (3:48 min). Available at:
[Link] Accessed on: 10 November 2023.

BURWELL, Catherine. The pedagogical potential of video remix: critical conversations about
culture, creativity and copyright. Journal of Adolescent & Adult Literacy, v. 57, n. 3, p.
205-213, Nov. 2013.

BUZAN, Tony. The history of mind mapping. Mind-Map - The World of Mind Mapping,
2024. Available at: [Link] Accessed on: 11 July 2024.

CASANAVE, Christine Pearson. Performing expertise in doctoral dissertations: thoughts on a

fundamental dilemma facing doctoral students and their supervisors. Journal of Second
Language Writing, v. 43, n. 1, p. 57-62, 2019.
 143

CASCONE, Kim. The aesthetics of failure: “post-digital” tendencies in contemporary

computer music. Computer Music Journal, Winter, Massachusetts Institute of Technology,
v. 4, n. 24 p. 12-18, 2000.

CHAKRABORTY, Utsho; GHEEWALA, Jaydeep; DEGADWALA, Sheshang; VYAS,

Dhairya; SONI, Mukesh. Safeguarding authenticity in text with BERT-powered detection of
AI-generated content. In: INTERNATIONAL CONFERENCE ON INVENTIVE
COMPUTATION TECHNOLOGIES (ICICT), 7., 2024, Lalitpur, Nepal. Proceedings […].
Xplore Part Number: CFP24F70-ART; ISBN: 979-8-3503-5929-9. [S. l.: s. n.], 2024. p. 34-
37. DOI: 10.1109/ICICT60155.2024.10544590.

CHRISTENSEN, Clayton. M. O dilema da inovação: quando as novas tecnologias levam

empresas ao fracasso. São Paulo: M. Books do Brasil, 2012.

COPE, Bill; Kalantzis, Mary; Searsmith, Duane. Artificial intelligence for education:
knowledge and its assessment in AI-enabled learning ecologies. Educational, Philosophy
and Theory, p. 1-17, 2020. Available at: [Link]

COSTA, King; MFOLO, Letlhogonolo Nolo; NTSOBI, Mfanelo Patrick. Challenges, benefits
and recommendations for using generative artificial intelligence in academic writing – a case
of ChatGPT. MetaArXiv Preprints, 2024. DOI: 10.31222/[Link]/7hr5v. Available at:
[Link] Accessed on: 11 Aug. 2024.

CRAMER, Florian. What is “post-digital”? In: BERRY, D.; DIETER, M. (ed.). Post-digital
aesthetics: art, computation and design. New York, NY: Palgrave Macmillan, 2015.

ČRČEK, N.; PATEKAR, J. Writing with AI: university students’ use of ChatGPT. Journal of
Language and Education, v. 9, n. 4, p. 128-138, 2023. Available at:
[Link] Accessed on: 4 Aug. 2024.

CROMPTON, Helen; BURKE, Diane. Artificial intelligence in higher education: the state of
the field. International Journal of Educational Technology in Higher Education, v. 20, n.
1, p. 1-22, 2023. DOI: 10.1186/s41239-023-00392-8.

DAVISON, R. M.; LAUMER, S.; TARAFDAR, M.; WONG, L. H. M. Pickled eggs:

generative AI as research assistant or co-author? Information Systems Journal, v. 33, p.
989-994, 2023. DOI: 10.1111/isj.12455.

DAWOOD, Mina Eshaq Tawfilis. Using artificial intelligence for enhancing human
creativity. Journal of Art, Design and Music, 28 June 2023.

DE OLIVEIRA, Luciana C.; SILVA, Tony. Second language writing in elementary

classrooms: an overview of issues. In: DE OLIVEIRA, L. C.; SILVA, T. (ed.). Second
language writing in elementary classrooms: instructional issues, content-area writing, and
teacher education. New York, NY: Palgrave Macmillan, 2016. p. 1-10.

DIBBELL, Julian (ed.). The best technology writing 2010. New Haven: Yale University
Press, 2010. 288 p.

DONG, Yifan. Revolutionizing academic English writing through AI-powered pedagogy:

practical exploration of teaching process and assessment. Journal of Higher Education
Research, Dalian, v. 4, n. 2, p. 52-57, 2023. DOI: 10.32629/jher.v4i2.1188.
 144

DOS SANTOS, Allessandra Elisabeth. Representações de professoras de inglês sobre

tecnologias e sua relação com a prática docente. 2020. Dissertação (Mestrado em Educação) –
Universidade Federal de Sergipe, São Cristóvão, 2020.

DOS SANTOS, Allessandra Elisabeth; OLESOVA, Larisa; VICENTINI, Cristiane; DE

OLIVEIRA, Luciana C. ChatGPT in ELT: writing affordances and activities. TESOL
Connections, May 2023.
[Link]
05-01/[Link].

DOS SANTOS, Allessandra Elisabeth; SILVA, Laila Gardênia Viana; SIQUEIRA, Nadege
Ramalho de. ChatGPT e implicações na educação: o cenário da escrita acadêmica. In:
PORTO, Cristiane; CHAGAS, Alexandre; OLIVEIRA, Kaio Eduardo (org.). Educiber:
educação e divulgação científica em plataformas digitais. 1. ed. Aracaju: Edunit, 2023, v. 5, p.
204-222. [Link]

DOUNG, Michael Tran; RAUSCHECKER, Andreas M.; MOHAN, Suyash. Diverse

applications of artificial intelligence in neuroradiology. Neuroimag Clin N Am, v. 30, n. 4, p.
505-516, Sept. 2020. Available at: [Link] Accessed on: 12
Nov. 2020.

DUNCANSON, Julia. Supporting L2 writing using QuillBot. GATESOL Journal, v. 33, n.

1, p. 30-36, 2024. ISSN: 2331-6845. Available at: [Link]
Accessed on: 4 Aug. 2024.

DUPPS, William J. Artificial intelligence and academic publishing. Journal of Cataract and
Refractive Surgery, v. 49, n. 7, p. 655-656, 2023. DOI: 10.1097/[Link].0000000000001223.

ELLAWAY, Rachel; TOLSGAARD, Martin. Artificial scholarship: LLMs in health

professions education research. Advances in Health Sciences Education, v. 28, p. 659-664,
2023. DOI: 10.1007/s10459-023-10257-4.

ELLIS, Amanda R.; SLADE, Emily. A new era of learning: considerations for ChatGPT as a
tool to enhance statistics and data science education. Journal of Statistics and Data Science
Education, v. 31, p. 128-133, 2023. DOI: 10.1080/26939169.2023.2223609.

ELSEN-ROONEY, Michael. NYC education department blocks ChatGPT on school devices,

networks. Chalkbeat, Jan. 3, 2023. Available at:
[Link]
artificial-
intelligence/#:~:text=%E2%80%9CDue%20to%20concerns%20about%20negative,education
%20department%20spokesperson%20Jenna%20Lyle. Accessed on: 16 Oct. 2023.

EWE, Koh. The ultimate election year: all the elections around the world in 2024. Time, Dec.
28, 2023. Available at: [Link] Accessed on: 15
June 2024.

FACHIN: ‘É HORA DE DIZER BASTA À DESINFORMAÇÃO E AO POPULISMO

AUTORITÁRIO’. RBA _Rede Brasil Atual, 18 July 2022. Politics. Available at:
[Link]
autoritario/. Accessed on: 15 Sept. 2023.
 145

FERRARI, Fabricio; CECHINEL, Cristian. Introdução a algoritmos e programação.

Material didático e instrucional. Porto Alegre, Rio Grande do Sul: [s. n.], 2008.

FISCHER, Steven Roger. A history of writing. London: Reaktion Books, 2021. 392 p.

FLORIDA, Richard L. The rise of the creative class revisited. New York: Basic Books,
2012.

FREEMAN, Donald. Doing teacher-research: from inquiry to understanding. Canada:

Heinle & Heinle Publishers, 1998.

FREIRE, Paulo. Educação como prática da liberdade. [Education as the practice of

freedom]. São Paulo: Paz e Terra, 1985.

FREIRE, Paulo; FAUNDEZ, Antonop. Por uma pedagogia da pergunta. [Toward a

pedagogy of the question]. São Paulo: Paz e Terra, 1992.

GIL, Antonio C. Como classificar as pesquisas? In: GIL, A. Como elaborar projetos de
pesquisa. 4. ed. Atlas: São Paulo, 2002. Cap. 4, p. 42-50.

GILBURT, Iona. A machine in the loop: the peculiar intervention of artificial intelligence in
writer’s block. New Writing, 21 June 2023. p. 1-12. DOI: 10.1080/14790726.2023.2223176.

GLOBO. Canadense usa inteligência artificial para simular troca de mensagens com noiva
morta há 8 anos. In: Fantástico. [S. l.], 26 Sept. 2021. video (12m 12s). Available at:
[Link]
[Link]. Accessed on: 10 Jan. 2022.

GORDON, Andrew D.; NEGREANU, Carina; CAMBRONERO, José Pablo;

CHAKRAVARTHY, Rasika; DROSOS, Ian; FANG, Hao; MITRA, Bhaskar;
RICHARDSON, Hannah; SARKAR, Advait; SIMMONS, Stephanie; WILLIAMS, Jack;
ZORN, Ben. Co-audit: tools to help humans double-check AI-generated content. [Link],
v. abs/2310.01297, 01 Oct. 2023. Available at: [Link] Accessed on:
10 Dec. 2023.

GUANGMING, Zhu et al. Artificial intelligence and stroke imaging: a west coast perspective.
Neuroimag Clin N Am, v. 30, n. 4, p. 479-492, Nov. 2020. Available at:
[Link] Accessed on: 11 Nov. 2020.

GUI, Jia; DOS SANTOS, Allessandra Elisabeth. Genre-based pedagogy in English language
arts integrated with generative artificial intelligence. In: DE OLIVEIRA, L. C. (ed.).
Applying a functional approach to language development in teaching the content areas
to multilingual learners in K-12 classrooms. Charlotte, NC: Information Age Publishing.
Forthcoming.

GUPTA, Rajiv et al. An east coast perspective on artificial intelligence and machine learning:
part 1: hemorrhagic stroke imaging and triage. Neuroimag Clin N Am, v. 30, n. 4, p. 479-
492, Nov. 2020. Available at: [Link] Accessed on: 11
Nov. 2020.

GUSTILO, Leah; ONG, Ethel; LAPINID, Minie Rose C. Algorithmically-driven writing and
academic integrity: exploring educators’ practices, perceptions, and policies in AI era.
 146

International Journal for Educational Integrity, v. 20, p. 1-43, 2024. DOI:

10.1007/s40979-024-00153-8.

HIGGINS, Eliot. Trump’s image arrest. 2023. Twitter: @eliothiggins. Available at:
[Link] Accessed on: 10 Nov. 2023.

HOSSEINI, Mohammad; RESNIK, David B. Guidance needed for using artificial intelligence
to screen journal submissions for misconduct. Research Ethics, OnlineFirst, 2024. DOI:
10.1177/17470161241254052.

JACOB, Sharin; TATE, Tamara P.; WARSCHAUER, Mark. Emergent AI-assisted discourse:
case study of a second language writer authoring with ChatGPT. [Link], v.
abs/2310.10903, 16 Oct. 2023. Available at: [Link]

JEANTET, Diane. Brazilian voters bombarded with misinformation before vote. AP News
headline, 25 Oct. 2022. Available at: [Link]
social-media-brazil-b75e51f515b0a708800198afa1b8601d. Accessed on: 13 July 2023.

KALANTZIS, Mary; COPE, Bill. Literacy in the time of artificial intelligence. EdArXiv
Preprints, Apr. 2024. DOI: 10.35542/[Link]/es5kb. Available at: [Link]
Accessed on: 10 May 2024.

KALANTZIS, Mary; COPE, Bill. Multiliteracies: a short update. International Journal of

Literacies, v. 30, n. 2, p. 1-15, 2023. DOI: 10.18848/2327-0136/CGP/v30i02/1-15.

KALANTZIS, Mary; COPE, Bill; PINHEIRO, Petrilson. Letramentos. Campinas, SP:

Editora da Unicamp, 2020. 408 p.

KAMMER, Michael N.; GOMILA, Paul; VUMBACO, David J.; MALDONADO, Fabien.
The role of artificial intelligence in scientific writing. Journal of Clinical Case Reports
Medical, 16 Feb. 2023. Available at: [Link]

KAUFMAN, Dora. A inteligência artificial irá suplantar a inteligência humana? Barueri,

SP: Estação das Letras e Cores, 2019. 94 p.

KAUFMAN, Dora; SANTAELLA, Lucia. O papel dos algoritmos de inteligência artificial

nas redes sociais. Revista Famecos, Porto Alegre, v. 27, p. 1-10, Jan.-Dec. 2020.

KIST, William. New literacies in action: teaching and learning in multiple media. New
York: Teachers College Press, 2005.

KNOBEL, Michele. Imagining beyond the domesticated “new”: creative remixings of

literacies, community, and place. An analytic review of the research literature. In:
LITERACY RESEARCH ASSOCIATION CONFERENCE SYMPOSIUM, 4 Dec. 2015.

KYRPA, Anna; STEPANENKO, Olena; ZINCHENKO, Viktoriia; DATSIUK, Tetiana;

KARPAN, Iryna; TILNIAK, Neonila. Artificial intelligence tools in teaching social and
humanitarian disciplines. Information Technologies and Learning Tools, v. 100, n. 2, p.
162-179, 2024. DOI: 10.33407/itlt.v100i2.5563. Available at:
[Link] Accessed on: 4 Aug. 2024.
 147

LANCASTER, Thomas. Artificial intelligence, text generation tools and ChatGPT – does
digital watermarking offer a solution? International Journal for Educational Integrity, v.
19, n. 10, 2023. Available at: [Link] Accessed on: 19
June 2024.

LEE, Kai-Fu. Inteligência artificial: como os robôs estão mudando a forma como amamos,
nos relacionamos, trabalhamos e vivemos. Tradução de Marcelo Barbão. Rio de Janeiro:
Globo Livros, 2019. 290 p.

LEMOS, André. Cibercultura: tecnologia e vida social na cultura contemporânea. 6. ed.

Porto Alegre: Sulina, 2013.

LEOPOLD, Seth S.; HADDAD, Fares S.; SANDELL, Linda J.; SWIONTKOWSKI, Marc.
Artificial intelligence applications and scholarly publication in orthopaedic surgery. The
Journal of Bone and Joint Surgery, v. 105, n. 11, p. 819-820, 7 June 2023. DOI:
10.2106/JBJS.23.00293.

LEU, Donald J.; SLOMP, David; ZAWILINSKI, Lisa; CORRIGAN, Julie A. Writing
research through a news literacies lens. In: MACARTHUR, C. A.; GRAHAM, S.;
FITZGERALD, J. (orgs.). Handbook of writing research. New York, NY: The Guilford
Press, 2016. p. 49-63.

LI, Xiaolian; ZHANG, Bo. AI poem case analysis: take ancient Chinese poems as an
example. In: CAIH2020: Proceedings of the 2020 Conference on Artificial Intelligence
and Healthcare, 4 Dec. 2020. p. 132-136. DOI: 10.1145/3433996.3434020.

LUCENA, Simone; SANTOS, Edméa. APP-DIÁRIO na formação de pesquisadores em

Programa de Pós-Graduação em Educação. Educação Unisinos, v. 24, n. 4, p. 658-671, Oct.-
Dec. 2019. Available at: [Link]

LÜDKE, Menga; ANDRÉ, Marli E. D. A. Pesquisa em educação: abordagens qualitativas.

São Paulo: Ed. Pedagógica e Universitária, 1986. Cap. 2, 3, p. 11-44.

LUKIANENKO, Valentyna Volodymyrivna; SHASTKO, Iryna Myronivna; KORBUT,

Oksana Hryhorivna. Evaluating AI detection tools for academic integrity in higher education.
Scientific Innovations and Advanced Technologies, v. 5, n. 33, 2024, p. 970-978. DOI:
10.52058/2786-5274-2024-5(33)-970-978.

MÁJOVSKÝ, Martin; KASAL, Matěj; KOMARC, Martin; NETUKA, David. Artificial

intelligence can generate fraudulent but authentic-looking scientific medical articles:
Pandora’s box has been opened. Journal of Medical Internet Research, v. 25, p. e46924,
2023. DOI: 10.2196/46924.

MALEKI et al. Machine learning algorithm validation: from essentials to advanced

applications and implications for regulatory certification and deployment. Neuroimag Clin N
Am, v. 30, n. 4, p. 433-445, Nov. 2020. Available at: https
[Link] Accessed on: 12 Nov. 2020.

MALIK, Agung Rinaldy; PRATIWI, Yuni; ANDAJANI, Kusubakti; NUMERTAYASA, I.

W.; SUHARTI, Sri; DARWIS, Arisa; MARZUKI. Exploring artificial intelligence in
academic essay: higher education student’s perspective. International Journal of
Educational Research Open, 2023. DOI: 10.1016/[Link].2023.100296.
 148

MCINTYRE, Lee. Post-truth. Cambridge, MA: MIT Press, 2018.

MCNAMARA et al. Automated evaluation of text and discourse with Coh-Metrix.

Cambridge, UK: Cambridge University Press, 2014.

MISRA, Durga Prasanna; CHANDWAR, Kunal. ChatGPT, artificial intelligence and

scientific writing: what authors, peer reviewers and editors should know. Journal of the
Royal College of Physicians of Edinburgh, v. 53, p. 90-93, 2023. DOI:
10.1177/14782715231181023.

MORIN, Edgar. Da necessidade de um pensamento complexo. In: MARTINS, Francisco

Menezes; SILVA, Juremir Machado da (orgs.). Para navegar no século XXI: tecnologias do
imaginário e cibercultura. 2. ed. Porto Alegre, RS: Sulina/Edipucrs, 2000. p. 19-42.

MORIN, Edgar. Introdução ao pensamento complexo. 5. ed. Porto Alegre: Sulina, 2015.
120 p.

MORIN, Edgar; ROGER-CIURANA, Emilio; MOTTA, Raul Domingo. Educar na era

planetária. Barcelona: Gedisa, 2003. 111 p.

MOURA, Francisco T. Artificial intelligence, creativity, and intentionality: the need for a
paradigm shift. Journal of Creative Behavior, v. 57, p. 336-338, 2023. DOI:
10.1002/jocb.585.

MUTHUKRISHNAN, Nikesh et al. Brief history of artificial intelligence. Neuroimaging

Clinics of North America, v. 30, n. 4, p. 393-399, Nov. 2018. Available
at: [Link] Accessed on: 11 Nov. 2020.

NEGUS, Keith; PICKERING, Michael. Creativity, communication and cultural value.

London: Sage, 2004.

NGUYEN, Andy; HONG, Yvonne; DANG, Belle; HUANG, Xiaoshan. Human-AI

collaboration patterns in AI-assisted academic writing. Studies in Higher Education, 2024b.
DOI: 10.1080/03075079.2024.2323593.

NGUYEN, Andy; ILESANMI, Faith; DANG, Belle; VUORENMAA, Eija; JÄRVELÄ,

Sanna. Hybrid intelligence in academic writing: examining self-regulated learning patterns in
an AI-assisted writing task. Frontiers in Artificial Intelligence and Applications, 2024a.
DOI: 10.3233/faia240198.

NOY, Shakked; ZHANG, Whitney. Experimental evidence on the productivity effects of

generative artificial intelligence. Science, v. 381, p. 187, 2023. DOI:
10.1126/science.adh2586.

OERMANN, Eric K.; KONDZIOLKA, Douglas. On chatbots and generative artificial

intelligence. Neurosurgery, v. 92, n. 4, p. 665-666, 2023. DOI:
10.1227/neu.0000000000002415.

OPENAI (2023a). GPT-4. United States of America. Available at:

[Link] Accessed on: November 20, 2023.
 149

OPENAI (2023b). ChatGPT can now see, hear and speak. United States of America.
Available at: [Link] Accessed on:
November 20, 2023.

OPENAI (2023c). ChatGPT (GPT-4 model, Mar 23 version) [Large language model].
Retrieved June. 25, 2024, from: [Link]

OPENAI. Discovering and enacting the path to safe artificial general intelligence. United
States of America. Available at: [Link] Accessed on: March 4, 2020.

OPENAI. How OpenAI is approaching 2024 worldwide elections. United States of America.
Available at: [Link]
Accessed on: June 14, 2024.

OXFORD. Word of the year 2016. Oxford University Press, 18 Nov. 2016. Available at:
[Link]
year/?cc=br&lang=en&#:~:text=After%20much%20discussion%2C%20debate%2C%20and,t
o%20emotion%20and%20personal%20belief.%22. Accessed on: 14 Nov. 2023.

PALFREY, John; GASSER, Urs; SIMUN, Miriam; BARNES, Rosalie Fay. Youth, creativity,
and copyright in the digital age. International Journal of Learning and Media, v. 1, n. 2, p.
79-97, 2009.

PAN, Jian. AI-driven English language learning program and academic writing integrity in
the era of intelligent interface. English Language Teaching and Linguistics Studies, v. 6, n.
4, p. 120-120, 2024. DOI: 10.22158/eltls.v6n4p120.

PARAGUASSU, Lisandra; MARCELLO, Maria Carolina. Brazil pro-democracy

manifestos aim to curb Bolsonaro’s election threats. São Paulo, 11 Aug. 2022. Available
at: [Link]
bolsonaros-election-threats-2022-08-11/. Accessed on: 13 Sept. 2023.

PARISER, Eli. The filter bubble: what the internet is hiding from you. London: Penguin UK,
2011.

PBS. Vencedora do Prêmio Nobel da Paz, Maria Ressa, sobre liberdade de imprensa. [S. l.]:
[s. n.], 2021. Video (6m 19s). Available at: [Link]
1639175584/. Accessed on: 10 June 2022.

PENG, Kuan-Po; MAY, A. Crossing the rubicon? the future impact of artificial intelligence
on headache medicine. Primer on Cerebrovascular Diseases, v. 43, n. 4, p.
033310242311573-033310242311573, 2023. DOI: 10.1177/03331024231157379.

PETRAGLIA, Izabel. Edgar Morin: a educação e a complexidade do ser e do saber. 13. ed.
[S. l.]: Ed. Vozes, 2020.

PETRAGLIA, Izabel; MORAES, Cândida Moraes. Curso introdução ao pensamento

complexo e implicações educacionais. Video. [S. l.]: Centro de Estudos Edgar Morin, 2022a.
Available at: [Link] Accessed on: 2 Oct. 2023.
 150

PETRAGLIA, Izabel; MORAES, Cândida. Epistemologia da complexidade e a pesquisa

educacional. Video. [S. l.]: Centro de Estudos Edgar Morin, 2022b. Available at:
[Link] Accessed on: 10 Oct. 2023.

PINTO, Maria da Graça Lisboa Castro. Os meandros da escrita acadêmica: alguns recados aos
estudantes universitários. Linha d’água, São Paulo, v. 31, n. 1, p. 9-27, Jan.-Apr. 2018.
Available at: [Link]

PLATE, Daniel Robert; HUTSON, James. Augmented creativity: leveraging natural language
processing for creative writing. Art and Design Review, v. 10, n. 03, p. 376-388, 31 Dec.
2022.

POLONSKY, Michael Jay. Should artificial intelligent agents be your co-author? arguments
in favour, informed by ChatGPT. Australasian Marketing Journal (AMJ), v. 31, p. 91-96,
2023. DOI: 10.1177/14413582231167882.

PRADO, Luiz; GARCIA, Carolina Borin. ChatGPT: entre o fascínio e o temor pela
tecnologia. Jornal da USP, 17 Mar. 2023. Available at: [Link]
entre-o-fascinio-e-o-temor-pela-tecnologia/. Accessed on: 18 Mar. 2023.

RABBIANTY, Eva Nikmatul; AZIZAH, Siti; VIRDYNA, Nina Khayatul. AI in academic

writing: assessing current usage and future implications. Insania: Jurnal Pemikiran
Alternatif Kependidikan, v. 28, n. 1a, p. 14-35, 2023. DOI: 10.24090/insania.v28i1a.9278.

RADFORD, Alec; WU, Jeffrey; CHILD, Rewon; LUAN, David; AMODEI, Dario;
SUTSKEVER, Ilya. Language models are unsupervised multitask learners. OpenAI Blog, v.
1, n. 8, p. 1-24, 2019.

RAHAYU, Rahayu; WEDA, Sukardi; MULIATI, Muliati; VEGA, Nofvia De. Artificial
intelligence in writing instruction: a self-determination theory perspective. XLinguae, v. 17,
n. 1, p. 234-244, 2023. DOI: 10.18355/xl.2024.17.01.16.

RAPPLER. About Maria A. Ressa. [S. l.]: [s. n.], [2021?]. Available at:
[Link] Accessed on: 10 June. 2022.

RATCLIFFE, Rebecca. ‘The Punisher’: Rodrigo Duterte’s violent reign as Philippines

president to end. The Guardian, 28 June. 2022. Analysis. Available at:
[Link]
as-philippines-president-to-end. Accessed on: 01 July 2022.

RAVAGNOLI, Neiva Cristina da Silva Rego. A anomia em um curso de formação

contínua de professores de inglês da escola pública: um olhar complexo. 2017. 142 f.
Thesis (Doctorate in Applied Linguistics and Language Studies) – Postgraduate Program in
Applied Linguistics and Language Studies, Pontifical Catholic University of São Paulo, São
Paulo, 2017.

REISS-ANDERSEN, Beirit. Announcement of the 2021 Nobel Peace Prize, 2021. Video
(15 h 51 min). Available at:
[Link] Accessed on: 10
June 2022.
 151

REZA, Mohi; LAUNDRY, N.; MUSABIROV, Ilya; DUSHNIKU, Peter; YU, Zhi Yuan
Michael; MITTAL, Kashish; GROSSMAN, Tovi; LIUT, Michael; KUZMINYKH, Anastasia;
WILLIAMS, Joseph Jay. ABScribe: rapid exploration of multiple writing variations in
human-AI co-writing tasks using large language models. [Link], v. abs/2310.00117, 28
Sept. 2023. Available at: [Link] Accessed on: 27 Jan. 2024.

ROE, J.; RENANDYA, Willy A.; JACOBS, George M. A review of AI-powered writing tools
and their implications for academic integrity in the language classroom. Journal of English
and Applied Linguistics, v. 2, n. 1, 2023. DOI: 10.59588/2961-3094.1035.

ROSCOE, Rod D.; SALEHI, Shima; NIXON, Nia; WORSLEY, Marcelo; PIECH, Chris;
LUCKIN, Rose. Inclusion and equity as a paradigm shift for artificial intelligence in
education. In: OUYANG, Fan; JIAO, Ping; MCLAREN, Bruce M.; ALAVI, Amir H. (ed.).
Artificial intelligence in STEM education: the paradigmatic shifts in research, education,
and technology. 1. ed. [S. l.]: CRC Press, 2022. p. 359-374. DOI: 10.1201/9781003181187-
28.

RUSSEL, Stuart J., NORVIG, Peter. Artificial Intelligence: a modern approach. 4. ed. New
Jersey: Prentice Hall, 2021. 1136 p.

RUSSELL-PINSON, Lisa; HARRIS, M. Lynne. Anguish and anxiety, stress and strain:
attending to writers’ stress in the dissertation process. Journal of Second Language
Writing, v. 43, p. 63-71, 2019.

SANTAELLA, Lucia. A educação e o estado da arte das tecnologias digitais. In: SALES, M.
S. (org.). Tecnologias digitais, redes e educação: perspectivas contemporâneas. Salvador,
BA: Edufba, 2020. p. 149-163.

SANTAELLA, Lucia. A Pós-verdade é verdadeira ou falsa? Barueri, São Paulo: Estação

das Letras e Cores, 2019.

SANTAELLA, Lucia. Comunicação ubíqua: repercussões na cultura e na educação? São

Paulo: Paulus, 2013.

SANTOS, Bergston Luan; ARRUDA, Eucidio Pimenta. Dimensões da inteligência artificial

no contexto da educação contemporânea. Educação Unisinos, v. 13, n. 4, p. 725-741, Oct.-
Dec. 2019.

SCHUTTE, Nicola S.; MALOUFF, John M. Connections between curiosity, flow and
creativity. Short Communication. Personality and Individual Differences, v. 152, p. 1-3,
2020.

SELIM, Abir Sabry Mohamed. The transformative impact of AI-powered tools on academic
writing: perspectives of EFL university students. International Journal of English
Linguistics, 2024. DOI: 10.5539/ijel.v14n1p14.

SHUMAILOV, I.; SHUMAYLOV, Z.; ZHAO, Y. et al. AI models collapse when trained on
recursively generated data. Nature, v. 631, p. 755-759, 2024. DOI: 10.1038/s41586-024-
07566-y.
 152

SRBA, I.; MORO, R.; TOMLEIN, M.; PECHER, B.; SIMKO, J.; STEFANCOVA, E.;
KOMPAN, M.; HRCKOVA, A.; PODROUZEK, J.; GAVORNIK, A.; BIELIKOVA, M.
Auditing YouTube’s recommendation algorithm for misinformation filter bubbles. ACM
Transactions on Recommender Systems, v. 1, n. 1, p. 1-33, 2023. DOI: 10.1145/3568392.

STANZIONE. Provvedimento del 30 marzo 2023.

[Link]
Accessed on: 15 Nov. 2023.

STOKEL-WALKER, Chris. ChatGPT listed as author on research papers: many scientists

disapprove. Visual Education, v. 613, n. 7945, p. 620-621, 2023. DOI: 10.1038/d41586-023-
00107-z.

STORNAIUOLO, Amy; HIGGS, Jennifer; NICHOLS, T. Philip; LEBLANC, Robert Jean;

DE ROOCK, Roberto Santiago. The platformization of writing instruction: considering
educational equity in new learning ecologies. Review of Research in Education, v. 47, n. 1,
p. 311-359, 2023. DOI: 10.3102/0091732X241227431.

SYSOYEV, Pavel V.; FILATOV, Evgeny M. Cognitive-matrix analysis as a tool for

exploring cultural identity of a fiction author. Vestnik Moskovskogo universiteta. Seriâ 19,
Lingvistika i mežkulʹturnaâ kommunikaciâ, v. 27, n. 2, p. 38-54, 2024. DOI:
10.55959/msu-2074-1588-19-27-2-3.

TAMILSELVI, C.; EBENEZER, A. Shamila; DHANASAKKARAVARTHI, B.; DEVI, R.;

DHANWANTH, Batini; MOHANAPRAKASH, T. A. A comprehensive study of AI-powered
writing assistance tools and content spinning applications. In: INTERNATIONAL
CONFERENCE ON ELECTRONICS, COMMUNICATION AND AEROSPACE
TECHNOLOGY (ICECA), 7., 2023, Coimbatore. Proceedings [...]. IEEE Xplore Part
Number: CFP23J88-ART; ISBN: 979-8-3503-4060-0. [S. l.]: [s. n.], 2023.

TAULLI, Tom. Introdução à inteligência artificial: uma abordagem não técnica. 1. ed. São
Paulo: Novatec, 2020. 232 p.

THE NATIONAL COMMISSION ON WRITING. Writing: a ticket to work. or a ticket out:

a survey of business leaders. [S. l.]: [s. n.], 2004. Available at:
[Link]/prod_downloads/writingcom/[Link]

THOMPSON, Paul. Achieving a voice of authority in PhD theses. In: HYLAND, K.;
GUINDA, C. S. (ed.) Stance and voice in written academic genres. UK: Palgrave
MacMillan, 2012. p. 119-133.

TU, Joseph; HADAN, Hilda; WANG, D.; SGANDURRA, Sabrina A.; MOGAVI, Reza Hadi;
NACKE, Lennart E. Augmenting the author: exploring the potential of AI collaboration in
academic writing. arXiv:2404.16071, 2024. DOI: 10.48550/arxiv.2404.16071. Available at:
[Link] Accessed on: 21 June 2024.

VALLARINO, Diego. How worried should you be about AI disrupting elections? In: The
Economist. 31 Aug. 2023. Available at:
[Link]
elections-of-2024. Accessed on: 15 Nov. 2023.
 153

VIBERG, Olga; KIZILCEC, René F.; WISE, Alyssa Friend; JIVET, Ioana; DOWELL, Nia.
Advancing equity and inclusion in educational practices with AI-powered educational
decision support systems (AI-EDSS). British Journal of Educational Technology, 2024.
DOI: 10.1111/bjet.13507.

WANG, Chaoran. Exploring students’ generative AI-assisted writing processes: perceptions

and experiences from native and nonnative English speakers. Technology, Knowledge, and
Learning, 2024. DOI: 10.1007/s10758-024-09744-3.

WIKIPEDIA CONTRIBUTORS. English Wikipedia. Wikipedia, The Free Encyclopedia.

2021. Available
at: [Link]
Accessed on: 15 Sept. 2021.

XU, Yang’en. The double-edged sword of AI writing tools. SocArXiv Papers, 2024. DOI:
10.31235/[Link]/wnpd5.

YAN, Da. Impact of ChatGPT on learners in a L2 writing practicum: an exploratory

investigation. Education and Information Technologies, v. 28, p. 13943-13967, 2023.
Available at: [Link] Accessed on: 4 Aug. 2024.

ZHI, Zhonghua Wei Chang Wai Ke Za. Role of artificial intelligence in the diagnosis and
treatment of gastrointestinal diseases. Chin J Gastrointest Surg, v. 23, n. 1, p. 33-37, Jan.
2020. Available at: [Link] Acessed on: 13
Nov. 2020.

ZARAMELA, Luciana. GPT-3: será que a IA escritora mais poderosa do mundo substitui um
jornalista? In: Canaltech. 28 Jan. 2021. Available at: [Link]
artificial/gpt-3-sera-que-a-ia-escritora-mais-poderosa-do-mundo-substitui-um-jornalista-
178031/. Accessed on: 15 Nov. 2021.

ZORLONI, Luca. OpenAI blocks ChatGPT in Italy. Wired, 31 Mar. 2023. Available at:
[Link] Accessed on: 10 Apr. 2023.
 154

APPENDIX I – Research credibility

Scientific events and activities Where? When?

Artificial Intelligence in English Language Writing. Oral Brazil-UFS Apr. 28
communication at I Seminário de Autoavaliação (PPGED). Synchronously/ 2021
Remotely
Pre-qualifying panel during the Research Seminar I course Brazil-UFS Sep. 30
(PPGED), taught by Prof. Dr. Simone de Lucena Ferreira and Synchronously/ 2021
Prof. Dr. Paulo Boa Sorte. Remotely
Presentation of the research project to TECLA research group Brazil-UFS Nov. 8
members. Synchronously/ 2021
Remotely
Generative AI writing assistants and their importance in USA-Virginia Mar. 10
education. Poster presentation at VCU Research Symposium – Commonwealth 2023
School of Education. University (VCU)
ChatGPT in Education: can we predict what’s going to USA-VCU Apr. 11
happen? Do we have to? Lecture for VCU Teacher education Synchronously/ 2023
program – School of Education. Remotely
ChatGPT: potential, applications, and challenges in education Brazil-UFS Apr. 14
and research. Live session for Coordenação de Pesquisa - Synchronously/ 2023
POSGRAP/UFS conducted by Prof. Dr. Renata Ferreira Costa Remotely
Bonifácio, research coordinator. @copes-ufs
Task-Based Language Learning and ChatGPT: how can we Colombia Apr. 21
connect them? Lecture for Masters Program in Education at Universidad 2023
Facultad de Educación. Universidad El Bosque, Colombia. El Bosque
Synchronously/
Remotely
ChatGPT in Education: how does it exist beyond limits and Brazil UNINASSAU June 17
challenges? Talk at Roda de Mestres AI: limits and challenges Synchronously/ 2023
for education teacher education program at Centro Remotely
Universitário Maurício de Nassau.
ChatGPT has come to stay. Oral communication at VII Brazil-UFS Oct. 10
Seminário Formação de Professores e Ensino de Língua In-person 2023
Inglesa (SEFELI).
ChatGPT and how it can be integrated into our education. A USA Oct. 14
podcast hosted by Nabeeha Haque for the course: Cyborg Georgetown 2023
Anthropology. AI Specialist Podcast University
Artificial Intelligence in Research and Postgraduate studies. Brazil-UNCISAL Oct. 17
Lecture co-presented with Prof. Dr. Paulo Boa Sorte at the II Synchronously/ 2023
Congresso Norte Nordeste de Pesquisa e Pós-graduação; V Remotely
Encontro da Pós-graduação Stricto Sensu e o II Encontro da
Pós-graduação Lato Sensu at Universidade Estadual de
Ciências da Saúde de Alagoas.
Presentation of ongoing research to TECLA research members. Brazil-UFS Nov. 13
Synchronously/ 2023
Remotely
Generative AI: shaping the new era of writing and Brazil-UFSC Nov. 16
communication. Talk for Academy UFSC Program, Synchronously/ 2023
Universidade Federal de Santa Catarina (UFSC). Novembro Remotely
entre Elas
ChatGPT in Academic Writing: ‘to use or not to use’ – is it still Brazil-UFS Nov. 28
a question? Course presented at the 33º Encontro de Iniciação In-person 2023
Científica, Universidade Federal de Sergipe.
 155

Qualifying exam presented to the Graduate Program in Brazil-UFS Dec. 05

Education (PPGED), Universidade Federal de Sergipe. Hybrid 2023
ChatGPT: exploring applications, ethics, and implications in Brazil-UFS Dec. 13
academic writing. Talk at the Department of Modern In-person 2023
Languages, Universidade Federal de Sergipe.
Optimizing AI Text Generators for Multilingual Learners: the USA Jan. 11
art of crafting effective prompts. Talk at 2024 Dr. Elizabeth GATESOL 2024
Webb Professional Learning Series – Disciplinary Language & Synchronously/
Literacy Communities of Practice, Georgia Teachers of Remotely
English to Speakers of Other Languages (GATESOL).
Leveraging AI Text Generators to Enhance Mentor Text USA Mar. 22
Creation for Multilingual Learners. Talk co-presented with TESOL 2024
Prof. Dr. Luciana C. de Oliveira during SLW-CALL* Tampa, FL
Intersection Section, Reimagining multilingual writing through In-person
AI, at the TESOL* 2024 International Convention & Expo.
*SLW (Second Language Writing), CALL (Computer Assisted
Language Learning), TESOL (Teaching English to Speakers of
Other Languages).
Creating Custom Prompts with Generative AI to Face USA Mar. 27
Academic Writing Challenges. Workshop presented for Berry College 2024
Faculty members at Center for Teaching Excellence and In-person
Faculty Development (CTEFD), Berry College.
Source: created by the author (2023).
 156

APPENDIX II – Scientific publications