Nigel Caplan

In Defense of Human(e) Writing

In summer 2024, I gave a plenary address to the Consortium on Graduate Communication online Summer Institute. The theme was “Change, Adaptation, and Engagement.” Engagement in writing I can support: good writers have to engage with their texts, messages, and audiences. But change and adaptation? That’s trickier. I believe too much change is happening in education through the uncritical and ill-advised adoption of technology. But I do see a need for a change in my field: one which reverts to the essentially human nature of writing. Instead of reacting and adapting to technological and institutional changes, I used my talk to call for pedagogical adaptations that make our work, the work of graduate educators and writing teachers more broadly, both more human and humane. What follows is an edited version of that plenary, which I organized around the three windmills I find myself tilting at (one for many years, one since 2020, and one quite recently).

Is it a chatbot or a false dichotomy?

The latest bad trend in higher education and professional conferences appears to be workshops, sessions, and webinars with misleading titles like:

AI: genie in the bottle or devil in disguise?
Does generative AI enhance or subvert learning?
Is AI the best thing since sliced bread or an inevitable pathway to the destruction of the human race?

One of these was a workshop at my university, which I did not attend on principle, and one is made up. I’ll let you figure it out. Also what’s a genie doing in a bottle, and is that a good … oh, never mind.

These false dichotomies are unhelpful, superficial, and fundamentally anti-intellectual. They imply that there only two positions that we can take towards so-called “AI” products: blind acceptance or alarmist denial. Of course they strongly suggest that the structure of the event will be a brief acknowledgement of the rampant harms of “AI” technology followed by fawning subjugation.

I resist.

These false dichotomies need to be exploded by applying the “critical thinking” skills that my profession has claimed to valued. Instead of setting up straw-people and forcing educators to pick a corner, our universities, libraries, and learned societies need to broadcast the facts about “AI” products. They’re not even contested:

Corporations that created large-language/content models stole copyrighted data.
The training and use of these products causes catastrophic environmental damage.
The output of “generative AI” models is racist (repeatedly).
The output of “generative AI” models is inaccurate (repeatedly).
The actual harms of “AI” products are real (corrupted medical data, dangerous fake medical advice, teenage suicide, fabricated publications, election interference). Meanwhile, any putative benefits are speculation bordering on fantasy.

To debate whether patently harmful technologies have a silver lining is absurd. These both-sides events show extraordinary disrespect for those who are vulnerable to the marginalization, racism, and pollution that tech corporations have chosen to inflict in the pursuit of power and profit. We all deserve better.

Today in “AI” exploitation …

The goals of so-called “AI” products and the massive tech corps that power them aren’t exactly subtle. I just can’t imagine the act of willful self-delusion it takes to believe that it’s OK to experiment with or encourage the use of ChatGPT and its ilk in education. This stuff is right there:

The Big AI Lie: “Never mind the quality, feel the width”

Last week, I published an editorial in TESOL Connections where I tried to inject some realism into the hype over so-called “generative AI” in language education. Many professional associations, including TESOL, have been promoting blog posts and workshops that encourage members to experiment with and incorporate these “AI” products in their teaching, and while I acknowledge the good intent behind these events, I wanted in my article to clearly set out the risks of these choices. And let’s be clear, the use of “AI” is a choice not an inevitability or requirement.

One type of response to my argument boils down to this: OK, I see what you mean about copyright infringement, invasion of privacy, bias, environmental impact, deskilling of the teaching profession, plagiarism, and misinformation … but look at all the social benefits AI will bring! As the saying goes, never mind the quality, feel the width.

Resisting “AI” and Refocusing on the Human

Here’s my op-ed column in February’s TESOL Connections, where I argue that:

What distinguishes our field is the human connection, something which AI technology can only supplant, not supplement.

I hope this is a useful corrective to the AI hype we’re hearing too loudly in education!

“Certified Human”: the new organic

One of the mitigations against the infiltration of synthetically-generated (“AI”) content is watermarking: the idea that text, images, video, and other content created with/by large-language models and other technology can be indelibly labeled as such. It’s not clear whether this is technically possible at the moment, and I can’t see how it’s desirable for the technology corporations promoting “AI”: after all, part of their hype is that such content is indistinguishable from the real thing, so why would they voluntarily slap a great big warning label on it?

In fact, I suspect the opposite will happen: creators will start putting a “certified-human” label on their work. I certainly plan to make it clear that my books, websites, lessons, syllabi, articles, workshops, and other content were not machine generated. My students, readers, editors, reviewers, and clients have a right to know that I do my own work and decide whether that’s something they value.* Think of it as the new organic.

Words matter: An “AI” primer

Technology corporations and their boosters are trying to shape perceptions of their products using obfuscating language. Unfortunately, this is getting picked up in causal use and especially in the media (et tu, BBC Newsround?).

So based on my reading and the tweets (or whatever) of people who know much more than me, here’s a quick list of DOs and DON’Ts:

Dear Student: Make the right choice, and keep away from ChatGPT*

* and all automated text generators or “help me write”/”magic” pop-ups

I live in Delaware not under a rock, so I’ve been trying to find a way to explain to my students that yes, I know what these so-called “AI” text generators are; yes, I understand that they are tempting; and no, I don’t want my students to use them. We can’t just pretend these things don’t exist, and it will be up to others not us to regulate and restrict them. I also don’t believe in surveilling my students or threatening them with grades (in fact, I’m ungrading). I’m an educator, so I want to educate.

By way of context, because all these decisions are situated, I teach at the university level, mostly courses that lead to a Master’s degree or teaching certification. However, I am also an ESL teacher and the chair of the instructional technology committee for our intensive English program. I don’t speak for that program, but I will be sharing these ideas with my colleagues.

I threw out an early draft of these guidelines on Twitter, and they spread far and wide (25K views and counting!). So here’s a more considered and better cited version.

Questions to Ask before Using AI in Education

These are the questions I’m asking myself, and which I’d like other educators, instructional designers, AI enthusiasts, and ed-tech promoters to ask

I can’t open my social media feeds or email these days without seeing another article about teachers using generative AI products (ChatGPT et al.) or an ad for a book promising 100 AI prompts for educators. I’m skeptical and concerned about the implications of using these products as teachers or with students. These are the questions I’m asking myself, and which I’d like other educators, instructional designers, AI enthusiasts, and ed-tech promoters to ask, too:

Previously on “Resisting Generative AI”….

Three pieces this week to counter the dangerous, headlong, blindfolded rush into accepting commercial generative AI products as if they were inevitable (they’re not):

Professor Flunks All His Students After ChatGPT Falsely Claims It Wrote Their Papers – no, obviously, ChatGPT doesn’t know if it wrote your students’ papers. It’s not intelligent. It’s a parlor trick that predicts text on the fly. Think chattering monkeys with a short working memory. Hard to imagine cases like this won’t end up in a very interesting court case, but hats off to Rolling Stone for the best attributive phrase of the year: ” a campus rodeo instructor who also teaches agricultural classes.”
Moving slowly and fixing things – We should not rush headlong into using generative AI in classrooms – from the always excellent LSE Impact blog. Debunks the “but what about calculators?” false analogy ( “It’s like comparing an abacus and a quantum computer”) and raises ethical questions about bias, plagiarism (OpenAI’s not students’), and equity.
Different field, but: TV writer David Simon weighs in on the Writers Guild of America strike – Simon wrote “The Wire”, so maybe he knows a thing or two about his craft, and certainly more than Ari Shapiro, who for some reason decides to play the role of naive AI-bro: SHAPIRO: So would you ever agree to a contract that saw any role for AI at all? SIMON: No. I would not. SHAPIRO: Huh. SIMON: If that’s where this industry is going, it’s going to infantilize itself. We’re all going to be watching stuff we’ve watched before, only worse. (Substitute classroom materials for TV scripts in this exchange and think about whether we want our labor to be devalued and our teaching to stagnate like this.)

Just because a technology exists does not mean (a) we have to use it; (b) we have to use it in education; or (c) it will inevitably become part of daily life. Instead we need to educate ourselves and our students about the risks of this technology and the benefits of doing the actual work.