ESSAY TWP20231

In today's era, education is undergoing a transformation with the positive

contribution of technology. In this context, we cannot overlook the power of
Artificial Intelligence (AI) in enhancing the examination process. Today's
presentation will take us on a journey to explore the application of AI in the field
of examinations, where it is not only innovation but also a comprehensive solution
to numerous challenges.
With the rapid advancement of technology, the education system is facing
increasingly complex challenges. Among these challenges, exam cheating has
become prevalent, affecting the fairness and reliability of the assessment process.
Nowadays, high-tech cheating devices are widely available for sale. With
extremely sophisticated designs and compact sizes, they can be hidden in
inconspicuous places such as under hair, in ears, etc., making exam supervision
much more difficult.
Traditional pen-and-paper exams and conventional monitoring software no longer
meet the increasing demands for transparency, accuracy, and effectiveness. This
raises the issue of the need for a method to accurately detect those using these
cheating devices. (1)
In this context, the application of Artificial Intelligence opens up new avenues for
the examination process. The autonomy, learning capabilities, and analytical
abilities of AI can revolutionize how we conduct exams and effectively combat
cheating. The presentation will address these potentials and how we can harness
them to enhance the quality and fairness in education.
Not only a breakthrough in educational technology, but the application of AI in
examinations also holds significant implications for the comprehensive
development of the education system. It promises to create a more stringent,
transparent, and accurate testing environment while reducing the burden and
pressure on both students and teachers. In an era full of challenges, AI support
could be the key to a more advanced and equitable education system.

Traditional methods of detecting cheating in examinations have long relied on

direct supervision, monitoring through cameras, and the use of monitoring
software. To combat exam cheating, invigilators often seat candidates far apart,
administer multiple versions of exams, diversify questions, etc. However, these
measures only reduce the likelihood of communication between candidates. Those
who resort to special methods to disclose exam content to outsiders for assistance
can be challenging to detect and prevent. Therefore, although these methods have
become standards for many years, they also have both advantages and
disadvantages.

One of the main advantages of these methods is real-time observation, allowing

invigilators to detect any suspicious behavior immediately. Furthermore, they
enhance a sense of responsibility among students since they know they are being
monitored. However, this approach has limitations because invigilators may not
catch every cheating instance, especially if adept students conceal their actions
effectively.
The use of cameras to monitor exam rooms is another common practice. This
method provides a broader view of the testing environment and allows for
reviewing footage later if there are suspicions. Cameras can capture various
activities, but they may not detect subtle cheating behaviors. Additionally, relying
on recorded footage means that detection does not occur immediately, potentially
allowing cheating behavior to go undetected during the actual exam.

Technological advancements have led to the development of monitoring software

designed to detect patterns or abnormal behaviors during online exams. These
programs can track gestures, head, hand, and eye movements. They may flag
suspicious pauses or even analyze eye movements through webcams. However,
they can produce inaccurate results, misinterpreting innocent actions as signs of
cheating. Moreover, the effectiveness of monitoring software depends on stable
internet connections and the availability of necessary hardware, which may not be
accessible globally. (2)
Common Advantages: Traditional methods allow for immediate identification of
suspicious behavior during testing. The presence of invigilators or monitoring
devices serves as a deterrent, preventing potential cheaters. Direct supervision
involves human assessment, which can be beneficial for various situations.
Common Disadvantages: Traditional methods may not cover every nook and
cranny of the exam room or effectively monitor all students simultaneously.
Monitoring methods can be perceived as intrusive and may affect the psychology
of test-takers. Monitoring systems, whether based on humans or technology, can
produce false positive or negative results, affecting the accuracy of cheating
detection. Setting up and maintaining monitoring devices or deploying monitoring
software can consume significant resources in terms of time, effort, and cost.

In summary, while traditional anti-cheating methods still have advantages, their

limitations require more advanced and sophisticated approaches. For example,
using Artificial Intelligence (AI) to address the increasingly complex challenges in
examinations. With the development of AI, methods for detecting cheating are
emerging and improving. They complement traditional methods, supplementing
each other, and even replacing them. For example, facial recognition, behavior
analysis, etc., have provided many benefits in supporting cheating detection in
examinations. (3)
Facial Recognition:
One prevalent application of AI is facial recognition, aiding in the accurate
identification of students participating in exams. The advantages of this method
include its automated and swift nature, creating an accurate map of each student's
presence in the exam room. However, the downside is the susceptibility of the
recognition system to various factors, such as poor lighting or changes in the
appearance of students. (4)

Gesture and Behavior Analysis:

AI can also analyze gestures and behaviors exhibited by students during exams,
tracking mouse movements, keyboard inputs, or even analyzing eye movements
through webcams. The benefits of this method include the detection of subtle
gestures and behaviors that may go unnoticed, automating the monitoring process.
However, challenges may arise from the difficulty in accurately assessing the
significance of each action, raising questions about the privacy of students.(5)
( Thái Khắc Văn – 20214145)
While AI brings substantial improvements to exam cheating detection, careful
consideration is necessary to ensure fairness, accuracy, and respect for the privacy
of all participating students.
Benefits and limitations of applying AI in prevention cheating in exams.
Benefits
Observe and alert teachers to students who are engaging in cheating quickly
and easily.
Not only can it be used in online exams, but it can also be integrated into
live exam tracking devices.
Allows students to participate and remain supervised even when not at the
test location.
Automated exam proctoring software helps reduce cheating in exams
conducted online - a must in the context of the ongoing Covid-19 pandemic.
Fairly organized exams help protect the value of certificates and degrees
awarded to students and ensure academic integrity. This is a key factor
required in awarding and granting certificates and degrees, especially in
some specific professional professions that require high ethics such as
medicine, pharmacy and law.
Besides, some automatic test proctoring software also can intervene on
student’s computers to block some specific software, tasks and monitor
suspicious actions. These software often use artificial intelligence
technology to monitor student’s exam attitudes. (tapchigiaoduc, 2021)
An abnormal behavior detection system applied to the problem of detecting
exam cheating can support exam invigilators to easily find students with
abnormal behavior, helping to improve exam transparency and accurately
reflect candidate’s exam results after each exam. (moi.gov, 2023)
Exam proctor using automatic exam monitoring software. (hoahoctro, 2023)
Normally, when proctoring an exam, each exam room has to have 2 proctors
to proctor the exam, but now there is an exam proctoring system that makes
proctoring easier.
Through surveys, the system detects cheating behavior with high accuracy,
over 97% with obvious behaviors such as turning sideways, turning face up,
lift up, glancing at cards, working separately under the table and the ability
to recognize Detecting the faces of candidates who commit cheating also
gives a very high accuracy of up to 98%.
 Exam anti-cheating system (tapchigiaoduc, 2021)
=> The use of artificial intelligence technology in exam prevention brings a lot of
convenience and accuracy to the learning environment. Teachers can more easily
proctor exams and accurately assess student’s abilities through tests.
Limitations
Requiring students to install software during online exams that requires too much
control over their computers is a security risk. In some cases, "remnants" of These
software remain even after students remove it from their devices.
Some students may not be able to purchase, rent, borrow, or have access to
technological devices or a strong enough Internet connection required by the
software. This can lead to technical errors during the exam, putting students at a
disadvantage and stressed. In one case surveyed by the research team, up to 41% of
students said they had encountered technical problems.
Exam proctoring software itself is also questioned about its ability to ensure
fairness. The facial recognition algorithms used in the software do not always work
correctly. Some research show that algorithms built by major software
manufacturers in the United States cannot recognize dark-skinned faces as
accurately as light-skinned faces. This subtle “discrimination” can be a sign of
underlying social inequality. Some other research have also raised concerns about
exam proctoring software in particular as well as facial recognition software in
general in this aspect.
Another issue is that test monitoring algorithms may mistakenly mark some of the
eye and head movements of students while taking the test. This can lead to
unfounded suspicions for some students who have neurological abnormalities or
simply have a different sitting "style" than the majority. Even without the influence
of automated software, exams themselves are very stressful for most of us and can
lead to some unwanted effects. (tapchigiaoduc, 2021)
EVALUATION OF CURRENT AND FUTURE TECHNOLOGY
Evaluate current technology
Some great achievements that AI technology has achieved to date:
Voice recognition: Voice recognition evolves in 2020 with higher accuracy and
better user experience. Some of the challenges that need to be overcome in this
area include:
 Access to more types of voices.
 Improve ability to recognize specific words.
 Minimize limitations on poor recording devices.
 Eliminate background noise.
 Video Analytics and Computer Vision.
Advances in AI/Deep Learning make image and video analysis more feasible at
low cost. Computer vision technology is greatly influencing the manufacturing
industry in introducing automated quality control and safety management
techniques.
Face recognition: Image processing and object recognition are one of the great
achievements in AI in 2020. AI applications for facial authentication are gradually
becoming familiar to smartphone users. Facebook and Google are the leaders in
developing facial recognition and object search.
 Face recognition technology (htt1)
Using artificial capabilities to create a monitoring system with high accuracy
during the exam process, not only applies to live exam monitoring but can
also be integrated into online exams. Furthermore, the software also supports
automatic grading. All of this helps reduce not only the teacher's effort but
also becomes a useful product for society, contributing to digital
transformation in education.
The more perfect this technology is, the more it becomes possible to
evaluate a person's capabilities, ensuring fairness and transparency. Artificial
intelligence does not have emotions, so it will give the most fair and
convincing results, minimizing the dependence of today's youth.

Identify the future of technology

In the future, the artificial intelligence industry is aiming to:
Computer vision: can contain one or more different intelligence techniques.
AI Application field: is the use of functional artificial intelligence techniques
or applications in specific fields and industries such as transportation,
agriculture, life sciences, medicine
 Artificial intelligence is the understanding and continuous development of
computer systems.
With predictions about the application of AI technology in many different
fields, researchers, businesses, startups and governments can guide future
development goals:
Support for government information portal systems: With the huge amount
of data stored in government databases, AI has the ability to organize and
combine multiple data sets to extract information. However, for effective
implementation, network security and information security issues also need
to be paid attention to and patched by the AI system.
Face recognition: Face recognition using AI is the application of artificial
intelligence to verify facial characteristics, computers automatically identify
and identify someone from a digital image or a frame in the video. Face
recognition using AI technology belongs to a branch of computer vision, in
which computer vision is much better than the human eye. One of the ways
to determine is based on the nodal points of the face. AI technology can
measure up to 80 nodal points (the distance between points on a face) to
make facial recognition (FR) easier.

AI technology can measure up to 80 nodal points (the distance between points on a

face) to make facial recognition (FR) easier. (htt2)
Currently, this application is integrated with many monitoring systems such as at
the company's timekeeping portal, monitoring systems at airports, train stations,
and public places; facial recognition in national security systems; in the security
system of banks, buildings... (hhtp.gov, 2021)
Technology is gradually being improved and developed in the future. Detection of
people showing signs of fraud will be evaluated by the machine through statistical
analysis of algorithms. Currently, the technology is being researched and still
requires humans. Each exam room will have up to 2 or 3 supervisors, but it is not
100% certain that everyone will be detected. In this regard, artificial intelligence
can completely replace humans. People who show signs of cheating or have
cheated will be transferred by the system to the server in the most accurate way.
From there, we can optimally reduce the number of human resources and still get
the most accurate results.
Applying AI in detecting exam cheating is an effective solution that can help
improve the efficiency of inspection and exam supervision, creating a fair and
transparent exam environment. However, to apply AI results effectively, there
needs to be a reasonable distribution among scientists, educators and educational
management agencies. Applying AI in preventing exam cheating and in education
in general is currently a global trend, first applied in developed countries, gradually
becoming inevitable and widely on a worldwide scale.
Reality shows that applying artificial intelligence in education brings many
benefits to countries. Modern technology platforms have helped in many aspects
and improved the quality of education in many countries around the world.
Not out of the general trend, the Vietnamese government is also in the process of
focusing on research, investment, development and perfecting the process of
applying artificial intelligence in the education system in general and exam issues
in particular.
We hope that through this presentation, everyone can get an overview of the
application of artificial intelligence in deteting exam cheating.

References
(1). https://vnexpress.net/tag/gian-lan-thi-cu-743904

(2). https://flyer.vn/phuong-phap-ngan-chan-gian-lan-trong-thi-cu

(3). https://laodong.vn/cong-nghe/sinh-vien-ung-dung-ai-phat-trien-phan-mem-phat-hien-
gian-lan-thi-cu-993347.ldo
(4). https://en.wikipedia.org/wiki/Facial_recognition_system
(5). https://www.cbim.rutgers.edu/research/behavior-and-gesture-analysis

(n.d.).

(n.d.). Retrieved from https://vietnamnet.vn/sinh-vien-lam-he-thong-phat-hien-gian-lan-thi-cu-nho-

cong-nghe-ai-758287.html.

(n.d.). Retrieved from https://www.poker.org/clearview-ai-facial-recognition-technology-may-be-illegal-

in-europe/

(n.d.). Retrieved from https://hauteliving.com/hautelawyer/635629/facial-recognition-technology-thing-

of-the-past/

(2021, 9 12). Retrieved from hhtp.gov: http://hhtp.gov.vn/vi/tin-tuc/cong-nghe-ai-cua-hien-tai-va-tuong-

lai-2128.html

(2023, 1 2). Retrieved from moi.gov: https://moit.gov.vn/tin-tuc/phat-trien-nguon-nhan-luc/truong-dai-

hoc-cong-nghiep-ha-noi-nghien-cuu-ung-dung-cho-ba.html

(2023, 1). Retrieved from hoahoctro: https://hoahoctro.tienphong.vn/thi-online-tai-cac-truong-dai-hoc-

giam-sat-chat-che-nhieu-ap-luc-khong-kem-thi-truc-tiep-post1389918.tpo

tapchigiaoduc. (2021, november 8). Retrieved from https://tapchigiaoduc.edu.vn/article/87200/225/cac-

truong-dai-hoc-dang-sudung-tri-tue-nhan-tao-de-ho-tro-viec-coi-thi-nhung-van-co-nhung-bat-
cap/?fbclid=IwAR1GArujv_vvhfTaLRCI0mzcDxa4x5EjC1nLMVRAoNQ1gD1seLapdgonOU4

(n.d.). Retrieved from https://nld.com.vn/giao-duc-khoa-hoc/kinh-nghiem-nhan-biet-gian-lan-thi-cu-

20230604200902191.htm

(n.d.). Retrieved from https://congnghiepcongnghecao.com.vn/tin-tuc/t24304/ung-dung-ai-phat-trien-

phan-mem-phat-hien-gian-lan-thi-cu.html

CAO VAN QUAN 20214057

THAI KHAC VAN 20214145
NGUYEN QUANG MINH 20203740