At the end of the term I asked my college creative wriing students to submit anonymous thoughts on AI. No real surprises: Mood ranges from resignation to despair, capitulation from embittered erosion of standards to total, feelings of betrayal from deep to furious. 1/

This is at Dartmouth College, where our president cut a deal with Anthropic--which stole the books of 133 faculty--without faculty consultation. Soon, there'll be a Claude for every student. They're already there. They feel like there's no escaping it. & they don't like it.

I wish I could tell you the responses thrummed with defiance. Instead, some read like substance abuse testimonies. A student whose use of AI grew until AI "wrote" all assignments sophomore fall. Got caught. Crushed by shame. Swore it off. But it's creeping back, & they don't know how to stop. 3/

(To be clear: I asked for anonymous responses so students could speak candidly & I could share comments.) A number of students wrote of their hate for AI, their disdain for friends who use it to do all their work, & then they said they "only" use it summarize readings or do research. 4/

Nobody used the term "arms race," but that's what they're describing: enormous pressure, including from some professors, to use AI or fall behind those who are. Many say they hate it, don't want to use it, but they feel like now it's submit or fail. 5/

Another student wrote with deep fury about AI taking over their education and wrecking mental health. They used to not touch it. Now, they use it to "supplement" their work in econ & math classes. They feel betrayed by the college and their teachers and themselves. 6/

I just learned that the term "carbon footprint" was developed by BP to offload responsibility for climate change onto individuals. Something like that's happening in higher ed with AI. Institutions capitulate or dodge; students are submitting; & they believe it's their individual failing. 7/

One student wrote of the belief among their classmates in another course that their professor had turned to "robograding," outsourcing evaluation of their work to AI. So why not use AI to write it? This student knows how cynical that is. It doesn't feel good. They wish there was another way. 8/

Several students wrote of the moral confusion implicit in a college that promotes AI and says don't cheat but makes it really, really easy to cheat. One said they don't pay attention to AI policies because they don't think the college really wants to know what's going on. 9/

One student, who wrote of trying and sometimes succeeding at cutting AI out of their life, wrote of watching a parent who is a lawyer turn over all their legal writing to AI. Almost every student wrote of how much AI is present in everyone's love lives. 10/

In winter, I spoke at a faculty event about AI. I was alarmed by the range of faculty awareness. Some believe our students are so "good" they'd never cheat. Yes, cheating's wrong. But systems that explicitly or implicitly incentivize cheating--AI does both--will produce cheating. 11/

Such a system will create conditions in which nobody's clear on what counts as "cheating." Some students who use AI to summarize readings for them don't think of that as cheating. It's about downloading the knowledge, right? (Wrong.) Worse, some faculty are iffy about it. 12/

One student noted that grade inflation is so pervasive that ethics policies around cheating have become hollow. This student said the only real guidance they felt they have about AI use is their own moral compass. They feel abandoned. 13/

(The answer to grade inflation, btw, isn't "tougher" grading. It's recognizing that grading, especially letter grading--not a very old system--serves employers but not students. It incentivizes cheating & corruption. Other evaluative frameworks are possible.)

Some people are asking if there were any "AI optimists." Many students would say something about how AI can be valuable for some things. But I was asking how it effects their education. In my small sample, no, nobody really described it as *improving* their education.

I've had a couple students plausibly report, anonymously, that they've never willingly gone near AI. (I say "willingly," because there are faculty who require it.) They're outliers, &, in their responses, part of that tiny cohort that's unaware of & immune to peer pressure.

There are other students who write of being able to imagine how AI could be valuable, or even interesting uses they've encountered in, say, computer science courses. But they express real dismay, to the point of despair, over the corporate complicity & enviro degradation involved.

A number of older students say they're glad they're getting out of college now, and they feel sorry for younger siblings and students for whom AI is so much more crushing that they'll have far fewer experiences of learning on their own.

Interestingly, nobody wrote about deepfakes and the problem created by AI of knowing what's real. I'm not sure if this is a good sign or a very bad one, evidence that "real" is less and less of a top concern.

I asked those who wanted to comment to print out their thoughts--no identifying handwriting, which, yeah, I get the irony--and then I'd leave the room so they could put them in a pile, or they cld leave them at my office. I really didn't want to know who wrote what.

I was reminded of Stephen King's comments on writing being "a meeting of minds on the page" and how AI cannot do this because there is no mind to meet. The writing is evidence of a process, of a presence, no matter how imperfect. One mind reaching out towards another.

That's what the feeling is when I read LLM generated text, of my brain slipping off the words and not being able to hold on. There isn't a person there.

One must wonder what becomes of minds that grow accustomed to LLM generated text as "writing"...when it's the old human writing that brains can no longer grasp

Do people read it? I thought they just summarised it using an LLM to gain a version doubly regressed to the mean of the actual writing it was trained on. In the end there is just one word and it's probably "utilise" or "agile".

I know people who only read AI summaries of books now. I.e. AI is the only "writing" they're reading. It's really nothing new: plenty of kids only read the Cliff's Notes for great books when I was in school 40 yrs ago. AI is more ubiquitous & accessible, therefore likely to become habitual.

I’m education faculty and I really appreciate the openness in which you address AI in your policy. I think I’m going to rewrite mine to include a statement that learning occurs through repeated practice with consistent feedback, and that gen AI removes those key components of learning.

I highly endorse ‘I'm not a cop’. I teach math at a community college (some remedial, nothing beyond a sophomore level), and some of the courses are online. Even before generative AI, if the students wanted to cheat, they could cheat. My job is to teach the students who don't want to cheat.

Sadly, our culture now promotes the idea that cheating is in fact the way to prosperity. It appears to be all about shortcuts and "hacks", not actually doing the work. Instant gratification without breaking a sweat.

Thank you. It's scary. Years ago, long before AI, I had trouble finding candidates for entry level jobs who had fairly basic comprehension and math skills. They were all college grads. About 16 years ago, I took a small business course with working adults - same problems. I worry for the future.

When I read your "complicated pleasure" in combining words and pictures, a photo you combined with an essay popped into my head. I didn't find that one, I found "Prayer Hands." No AI as an experiment in real. killingthebuddha.com/mag/icon/pra...

Hell to the yeah for moments of clarity and the thought of your mind racing..

I like how you appeal to their identity / sense of self and fulfillment in this. It's inspired me to try a similar angle with my incoming first-years in September.

I can no longer find the attribution, so it may be apocryphal, but I have long held dear a quote that I remember as coming from Arundhati Roy: "A sentence thrills me."

All the AI nonsense aside, I can’t tell you how excited I would’ve been to read a statement like that in a college class.

Please turn this into an essay or something that I can share with people who don't do Bluesky (or social media).

Ironic, eh. In 1990-1, I was at an Institute of Technology, bought a 2nd hand luggable, turned in my work printed, using the free Microsoft mini office sw. My work was rejected and it was demanded I did it by hand. I'd already done so, so that was easy. I wanted to use the new tech lol, 1)

...but I remember everything from those classes because i was given TIME to handwrite notes from 3-hour lecture slides, and we did all our prac work in groups helping each other. Changed institutions, plagiarism was excuse for no longer doing group work unless prescribed, suddenly, 2)

...nobody in the workplace knew how to cooperate, everyone kept their methods and mistakes secret, and ironically this had come in with the acceptance of computers to present work. Also trendy engineering recruitment practices led to selecting only certain personality types leading to stagnation! 3

I would take your course just based off this policy statement.

Thank you for recording this important gestalt and asking and receiving in the way that allowed for this frank insight. The kids are alright.

"I'm not a cop" works because it appeals to values rather than enforcement. The harder test is in required courses, where students may not care about writing for its own sake. Does the same approach work when writing isn't the main objective?

I got really emotional reading that AI policy. That is what writing is and does for me. I don't want anyone with that spark to abandon that complicated process for the simple, frictionless way out.

Thanks for sharing your policy, and for this entire thread. It's really helpful and thought-provoking as I work on my syllabi for the fall.

I enjoyed your statement. As a HS writing teacher, I’m coming closer to something similar: cheating is between the student and their gods. I’m redesigning some grading this summer to try and reduce the incentives to LLM assignments, and simultaneously implying i’ll just give C- and move on

I think this is particularly compelling and directly tailored for a writing course. Thanks for sharing it with others.

This is very interesting. As the first step, I would start calling things what they really are. There is no AI out there, only stochastic language models trained with a single goal - to maximize the profit of their developers.

This makes me wish I could take your creative writing course. You seem to find the same beauty in the process as I do, with my ever-growing pile of living documents that believe themselves stories... This makes me wish my engineering degree had afforded me space for a creative writing course.

Thank you for this whole thread. I want a better world for these students; I want them and those that come after them to know the joy of learning. I'm no longer optimistic but maybe we (that is higher ed) can find our way back there eventually.

Jeff, this made me cry

1) Before trying to suggest anything, what sort or rubric or grading scheme are you using? 2) College is insanely expensive, most leave with debt, and the job market sucks; no surprise students will do what they can to maximize success, and this thread makes clear they feel they have no choice.

It's great, but how do you feel knowing that much of the work you marked wasn't "real", and did you notice?

in most cases, a human can't notice, and anyone saying otherwise is puffing themselves at the expense of confronting a crisis. How does that feel? Bad.

Students have to learn to write AND think AND critically evaluate what they read. I am pondering how do this (although I don't have much teaching now), and the only thing I can think of is: handwritten exams and oral examinations, and discussions around an issue rather than essays.

Can’t help but think “using AI would be a complete waste of time and attention from a qualified, thoughtful expert in the field.” Some kids will still do it, but they are cheating themselves most of all.

What to even say? One thing to point out is that this degradation of education has been coming since the internet appeared. In the 2000s, I had students (& knew of more) who submitted essays that had been copied off the internet, either verbatim or were patched together from multiple sources

This makes for very depressing reading. I'm rather like the older grad students--I'm glad that my kids are close to the end of their educations so that they won't be as influenced by AI's detrimental effects.

Well isn’t that lovely. I bet you make a lot of students very happy when they read that.

Love the practical poetry.

_ > I am a student at Dartmouth college and we are asked to use Claude. However, I am not too happy with it but I can't figure exactly why. 1/4

_ reasoning style, academic usefulness 2/4

_ Poetic and flat 3/4

_ Whatever AI is producing, it is under the Prompter Guy signature and responsibility. I do not like, I do not sign. 4/4

Jeezus, I remember when students would be sternly reprimanded for plagiarizing Wikipedia & books. How much farther have we fallen. Now they want something much worse than plagiarism. Having students let a robot completely write for them... This timeline keeps on sucking in the worst ways.

I was thinking of myself, a first gen student who had no support for writing here. Not sure what level this course is, but some nod about welcoming those who are hesitant but ready to grapple could be useful for a hesitant student.

A lot of your observations are kinda like why we're getting "thinkpieces" about how young people don't want to work hard. It's not that they're unwilling to work hard, but only if it can pay out. Working twice or more harder where your only rewards is expectations of you working even harder in the

Not sure I follow. I said just the opposite. The students responding to me don't want these "shortcuts." For various reason, most use them anyway. And not everyone is making life choices based on predator-prey efficiency calculations. Nobody taking CW courses is, because efficient it's not.

They're using them despite not wanting to for various reasons like believing the grading uses AI. Not many people want to feel like their effort, especially extra effort is without purpose. They're probably there to learn, but the administration isn't focused on that to them.

Like your survey, you made it anonymous as possible in part so it felt safe to be honest. This stuff isn't math calculations, it can be unconscious. But most choices have reasons, especially when it's against your wants.

Great thread! Makes me glad I'm not in college having to deal with Ai. It's bad enough out in the wild trying to not get inundated in the Ai bs.

thank you very much for sharing this

Time for a blue book