Does Your Department Have an AI Policy? Here’s Edinburgh’s
Has your department instituted an AI policy? If so, whom does it govern, and what does it say? What should such a policy say?
Has your department considered an AI policy but held off on writing or implementing it? If so, what issues, disputes, or questions have contributed to the delay? Does your department even have the institutional authority to have such a policy? Would it be better to not have a policy?
These and other questions are prompted today by word that the Department of Philosophy at the University of Edinburgh recently adopted a policy for the use of AI by students in philosophy courses.
It’s the default policy, which means it can be superseded by policies developed by individual instructors for their courses or approved disability accommodations.
Here it is:
Policy on the use of AI in Philosophy Courses
Summary: Edinburgh philosophy is human made! The use of AI is prohibited when completing assessed work in philosophy.
Definition:
AI includes but is not limited to
-
- Generative AI systems capable of producing essays, summaries, explanations, or arguments
- AI systems that use Large Language Models (LLM)
- Automated paraphrasing or rewriting tools
- AI systems that generate images, audio, or video
- AI systems that answer questions or provide explanations in natural language
This definition includes all models of ChatGPT, all models of Gemini, all models of Claude, similar tools such as Midjourney, Microsoft Co-Pilot, and Grammarly, as well as all tools hosted on the ELM platform of the University of Edinburgh.
Note: Standard spell-checking software (e.g. in Microsoft Word), citation management software, and grammar checkers that conform to the university’s Proofreading Policy are excluded from the definition of AI.
Statement of policy: Philosophy courses aim to foster careful original thought, improve persuasive academic writing skills, and enhance students’ abilities to understand complex topics. Using generative AI tools to produce assessed work is almost always detrimental to this aim, which is why the use of AI is prohibited in writing, revising, and editing of academic work that is submitted for assessment in Philosophy. This includes undergraduate and postgraduate essays, undergraduate and postgraduate dissertations, take-home exams, or any other form of assessment.
Violations of the policy may be referred to the School Academic Misconduct Office (SAMO) for investigation.
Policy status: This is the default policy for all courses in philosophy, including dissertation courses. Course Organisers can replace the policy with a policy of their own. Also, approved learning adjustments take precedence over the policy.
A member of the department also shared the justification for the default “no AI use” policy for students:
Justification of policy:
-
- Educational significance: The articulation and organization of one’s thought through critical reading and linguistic expression is the very activity that constitutes the study and practice of philosophy. The use of AI in writing, revising, and editing academic work avoids engaging in the very activity that the study and practice of philosophy consist of. The use of AI directly hinders philosophical learning. (You wouldn’t send a robot to the gym and expect to get fit; it is the same with learning.)
- Academic integrity: The written work that is submitted for assessment in a course is a manifestation of a student’s learning in a course. The use of AI violates the principle of academic integrity since it represents something as one’s own learning that is not so. This is straightforwardly dishonest.
- Intellectual autonomy: In leaving the articulation and organization of thought to an AI system, one eschews an activity that is integral to the study and practice of philosophy. This is a failure of intellectual integrity: it is a failure to think for oneself.
- Cognitive harm: Philosophy aims todevelop your knowledge and teach certain skills, chiefly thinking skills. Using AI not only prevents the acquisition of knowledge and the development of thinking but undermines thinking skills you already have: ‘cognitive offloading’, ‘metacognitive laziness’ and reduction in capacity for critical thinkingare associated with this technology.
- Quality: Text generated by AI often looks good: it is well-structured, polished, and correctly formulated, and it uses technical terms in a way that makes it seem as if there is genuine understanding. However, AI-generated text is often shallow and stereotypical, and sometimes downright false. It looks better than it is.
- AI reproduces bias: AI systems that are trained on texts that are available online replicate the biases that are internal to these texts. AI systems also exhibit algorithmic biases, intended and unintended, none of which are visible to users of these systems.
- AI is piracy: Many AI systems are trained on texts that are available online. Most of these texts were not written with such training in mind, and consent from their authors was not obtained. AI systems are, therefore, built on piracy in the service of profit-driven corporations. Contributing our own prompts and documents to these systems amounts to collusion in such piracy.
- Environmental impact: AI systems use enormous amounts of energy and have a very large environmental impact, which exacerbates the climate crisis.
- Political significance: Most AI systems are produced by profit-driven corporations, even when they are sold to universities as separate systems. AI systems are marketed with the narrative that integration into higher education is inevitable, tapping into people’s anxieties about writing, thinking, and scholarship. However, this narrative is false, and the outcome is not inevitable: The corporate takeover of language and thought will only succeed if academics and students allow it.
Discussion welcome.
Here we go again, trotting out policies that sound like sticking one’s chin out and denying the future. We are in the age of AI now, like it or not. Policies like Edinburgh sound like the old days in the industrial revolution where lumberjacks like Paul Bunyan were against the invention of the chainsaw for cutting down trees – they wanted to insist it keep being done with an axe. Guess who went out of business (it wasn’t the chainsaw companies). Bunyan was in the age of the chainsaw, and Edinburgh University in 2026 is in the age of AI; the aim should be a reasonably AI integration policy not hamfisted denialism.
Well, chainsaws didn’t cause cognitive decay in the youth, although i suppose they did probably cut some people. Anyway though, your general point does seem right – which is that policies that involve hermetic seals like Edinburgh’s constitute a form of denying the age we are in. so i think you’re right that if we live in an AI age, banning AI in one’s education is goign to leave someone ill equipped to take on the challenges OF THE AGE, namely, an age deeply and inextricably linked to AI.
Um you do know that Paul Bunyan wasn’t a real person don’t you? I guess I should be happy you didn’t bring Ichabod Crane, Johnny Appleseed, or Captain James T Kirk into it.
Anyway over and above an amusing sloppiness with facts, the future is X so we must do X is an ambitiously awful sort of argument. You may as well argue that we should allow cars in the Boston Marathon or motorcyles in the Tour de France since we’re in the age of the internal combustion engine. We have a choice about what sort of future we want. We don’t have to give in to the tech bros and condemn our children and everyone else’s to unending AI slop and brain rot.
Not to mention, we don’t have to (and shouldn’t) simply accept the devastating environmental consequences of AI as inevitable. I think this important issue sometimes gets forgotten, when in my opinion it belongs at the top of the priority pile.
In the interest of pedantry, Johnny Appleseed was a real person.
Exactly!
Contrary to your claim, Paul Bunyan did exist! Paul Bunyan’s real name is Fabian Fournier; of course some of the tales of his deeds are of doubtful authenticity and origin. But Paul Bunyan existed if Fabian Fournier existed and the latter did (compare, Superman existed if Clark Kent existed). And more to the point of this discussion, Fournier used an axe, not a chainsaw.
You raise an excellent point. And I actually just found out that it turns out that Ichabod Crane exists too. He isn’t named Ichabod Crane, he’s actually a woman and he runs the local BP station instead of teaching school. And it turns out the headless horseman isn’t a Hessian or Brom Bones but a local kid who wanted to buy cigarettes without an ID.
What a strange thing to say about Ichabod Crane.
“We are in the AI age now, like it or not.”
We are similarly in the age of the nuclear bomb, like it or not. But would you agree that we should just throw up our hands and let that technology develop untrammeled and unregulated? We might not be able to un-invent a piece of tech once the proverbial genie is out of the bottle, but if that technology has the potential to be harmful (as AI certainly does), then it is perfectly reasonable to say that its proliferation should be subject to limitations and regulations preventing those harms. Maybe you think Edinburgh is going too far with this, but their policies seem aimed at making sure students can’t outsource their critical thinking to computers, which strikes me as laudable.
Bilingual, maybe the following gentle analogy will be of help: from a pedagogical point of view, we want to at least, and among other things, prepare students to make their way in the world. Can we agree on this? From this point, though, consider the foolhardiness of training students hermetically sealed from AI like, in the Allegory of the Cave, being hermetically sealed from the sunlight. When students learning without AI at all then make their way out into the world (now deep-integrated with AI), they will, as it were, be blinded by the sunlight, unfit for adapting, unable to cope.
I understand the point you’re advancing, though you haven’t really engaged with the thrust of mine vis-a-vis harms and regulations. Perhaps it would help to be more specific about what scenarios you have in mind that would require philosophy graduates to be familiar with AI systems. A recent DailyNous thread involved evidence of a negative correlation between AI use in particular disciplines and the career prospects of students graduating in those disciplines.
Peter
Arguably, the students who are working with AI are merely seeing shadows. Edinburgh is trying to help them see this, and take them out of the cave, into the sun. Of course, they (and many others) are not interested in this inevitably painful experience.
“AI exists” is a complete non-sequitur w.r.t. its role in education. The analogy that the policy notes, which I and may others have used ourselves, is using a machine to lift the weights in a gym. We live in the machine age, but using a forklift won’t make you strong. It’s not denialism about the machine age that at play here: if you’re stocking a warehouse, the forklift is a great tool. But if you’re trying to get strong, the forklift is stupid – more precisely, you’re confusing product and process.
DLC, this is actually a pretty artfully constructed trolling attempt, I award it 8/10. You managed to snag quite a few people. Next time, though, I suggest avoiding dead giveaways, like repeated reference to people who did not exist, or using really tired cliches. I think you’ll trick even more people into responding to you.
Compare these accusations of “denying the future” with the fact that, “when Central Park in New York was initially zoned as development-free, it was a worthless plot of land. Now, people pay great sums to have an apartment with a view on Central Park. You will of course diverge from current trends if you make your graded assignments AI-free. But, don’t worry: Delineating a space for trees in a time of rapid urban expansion wasn’t ‘backward-looking.’ On the contrary, it was VERY forward-looking” (taken from https://certifiedaifreeskillsandknowledge.org/).
That’s a beautiful analogy but people still want to live in their apartments, not in the park.
Good. This shows that “[m]aking your courses AI-free won’t ‘deprive’ anyone of anything since, outside of class, students remain able to type prompts in a chatbot. The reverse however does not hold, since those reliant on machines cannot do what truly skilled and knowledgeable people can do. The value of an AI-free education is thus bound to climb” (pasting again from https://certifiedaifreeskillsandknowledge.org/).
I don’t disagree! But your analogy involved not just the value of Central Park but the value of surrounding apartments. Presumably AI-open courses also gain value from AI-free courses according to the analogy?
So reasons 3, 7 and 9 would all seem to pretty straightforwardly rule out using Microsoft’s grammar checker.
Agree – I can only imagine here Edinburgh is tapping in to its glory days of the Scottish Enlightenment and hoping to go back to writing in quill pens on rough parchment.
And indeed, if their statue of Hume is any indication, the students will soon be wearing anachronstic greek robes to class.
There might be good reasons for completely banning LLMs, I don’t know. But some of the reasons, which are supposed to be reasons against using them as a grammar checker, would also apply to using just about any contemporary grammar checker, as far as I know.
9 would seem to be a reason against using any tool whatsoever.
Right. Students who are at a university level should do their own grammar checking.
Okay, but that’s not what the policy says. And the claim that using a grammar checker is an academic integrity violation is surprising and is definitely a departure from standard practice prior to AI.
First, all this is a definitely a departure from standard practice prior to AI.
What’s different with grammar checkers is that some, notably Grammarly, have now evolved to be much more than mere grammar checkers. So, it’s important to call them out as verboten to avoid any ambiguity. If you leave that door open even a crack, a flood of abuse may come pouring in.
Second, you may have missed this bit in the policy that directly addressed your concern:
I didn’t miss anything. That’s exactly what I was referring to. Grammar checkers definitely violate 7 and 9. And I can’t think of any reason why using an LLM to check grammar would be ruled out by 3 that isn’t also a reason why using a grammar checker would be ruled out by 3.
Well, one reason is what I already gave you:
Another reason may be that they didn’t want to depart from standard practice, as you were concerned about. So, they may have stipulated a carve-out for non-LLM grammar checkers, not necessarily for principled reasons but for the sake of not upsetting standard practice where they can help it?
Sorry, your claim is that using an LLM to check grammar is a failure to think for oneself, because it has other capabilities?
You’re also claiming that using grammar checkers of any sort would be considered an academic integrity violation because they are trained on large text corpuses and are made by for profit companies, except we’ve chosen to grandfather them in because everyone is so used to them?
They explicitly state that they only allow grammar checkers “that conform to the university’s Proofreading Policy”. To me this seems like a reasonable sort of policy – “grammar checkers” often do a lot more than just notice subject-verb agreement.
They do intend to rule out many grammar checkers! And I think this is reasonable. (Though I don’t think reasons 6-9 are great reasons for the policy.)
They’re inconsistent with the policy.
can you state how you see this to be the case?
“Standard spell-checking software (e.g. in Microsoft Word), citation management software, and grammar checkers that conform to the university’s Proofreading Policy are excluded from the definition of AI.”
it seems clear to me that 7 and 9 as they are worded address content rather than grammar. your interpretation of 3 would seem to suggest that grammar would count as “articulation and organization of thought.” while grammar surely plays
a role in articulation and organization of thought, it’s not at all clear to me
that standard grammatical errors like subject/verb agreement are meaningfully
central to either.
and if we note that grammar is at least in part arbitrary, and that those who are multilingual may commonly commit grammar errors that don’t have to do with organization but with arbitrary conventions (take for example how chinese speakers often struggle with articles in english [and vice versa, no doubt]), i’m not sure i see your case.
I’ve already explained these points. If using an LLM to check grammar violates 3, so does using a grammar checker to check grammar.
7 and 9 are not claims about content at all. They are claims about how the software is produced. 7 claims the entering a prompt into an LLM is collusion with piracy.
I already explained this. If using an LLM to check grammar violates 3, so does using a grammar checker. (Or if using an LLM to check grammar violates 1, if that’s the principle that’s more focus on the mechanics of writing, so does using a grammar checker.)
7 and 9 say nothing about content. They are both claims about the origins of the software. Grammar checkers are also trained on large amounts of text to detect statistical patterns in language. If this makes LLMs depend on piracy, grammar checkers depend on piracy as well. 7 states that merely entering a prompt into an LLM is collusion in piracy.
9 states as a reason for disallowing that LLMs are produced by for-profit companies. This should rule out, in the places most of us live, using grammar checkers, licensed software, computers, books, published articles, pens, pencils, paper, desks, chairs and so on. Presumably the only way to access many published articles without contributing to for-profit companies would be to pirate those articles, which would violate 7.
I agree we should note that grammar conventions are in part arbitrary. But I don’t really understand what that has to do with the issues I’m noting with these principles.
Sorry about the double-reply. The first one didn’t show up on my end until several hours later.
Hear hear!
This is one of the best AI policies I’ve seen in education. Finally!
It’s clear, brief, but also detailed enough to be substantial, persuasive, and a nice starting point for discussions.
And it’s flexible enough (in that it may be superseded by an individual instructor’s policies) to accommodate AI optimists and others who might want to experiment with LLMs.
Well done, Edinburgh philosophers!
A rare agreement between you and me on this topic!
🙂
I suppose the justification works as a kind of cumulative argument, for many of the reasons do not, on their own, justify such a restrictive policy. For mainly two reasons: some of the listed claims are contested (e.g., environmental costs, cognitive harm); and some overgeneralize (at least half of those claims would justify prohibiting using many resources which, while suboptimal for a variety of reasons, we consider acceptable). But I take it that, together, these claims do justify eschewing the use of AI by students. Which prompts two questions:
1. How will the policy be enforced? Presumably not using tools that fall prey to many of the objections listed under the justification. If not then, is the policy enforceable in its wide scope?
2. Is there discussion of extending it to apply to faculty too? If not, why not? If yes, why?
I think justifications 1-5 clearly motivate not using AI for most educational assignments, while not applying to research, teaching, or assessment. Justifications 6-9 seem to me like overreach, both relying on false estimates of the scale of the issue (and the comparable issues for traditional methods).
Enforcement is a difficult issue for any of these policies. But if you have a lot of practice reading AI writing, you can identify a lot of the tells, and as I suggested above, justifications 1-5 for students not using these tools to complete their assignments don’t obviously motivate a ban on using parallel tools to check whether students are violating the policy.
Justifications 1-4 are compelling to me and motivate my own AI policy. I’m on the fence about 5 because my impression is that the jury is still out on ‘cognitive harm’ (a loaded term) and much of what is established is already covered by 1 and 3, so it’s not doing any independent work other than using scary words on the basis of contested evidence. BUT I’m willing to grant that extensive use of AI is bad for students’ intellectual development in some sufficient sense. I agree about 6-9 and enforcement.
Sorry I meant 1-3 and 5 and I’m on the fence on 4.
I think this underestimates the difficulty. Enforcing such a rule isn’t as simple as noticing the clear AI tells and giving the student an F. That’s because some students will appeal this decision. And such appeals will become increasingly common, because students will soon figure out that we cannot prove that our suspicion of AI use is true – at least, not to the satisfaction of a dean or a provost whose eye is on the university’s bottom line. My bet is that administrators will not support the faculty member over the student in such cases. There have already been lawsuits by students who say they have been wrongly accused of using AI.
It will be genuinely interesting to see how this works out for the Edinburgh department.
Moreover, the justifications for the policy rule out using AI-detection software. I’ve been able to catch students without ever relying on such tools, but only because the students confessed.
I think this is an excellent policy. Of course since they provide explicit justifications, it is possible to raise objections to any or all of them (I think some are weak or irrelevant and probably counterproductive as well) but in offering critiques we should bear in mind that that the justifications offered by the AI boosters, which aren’t always made explicit, are not exactly airtight: “everyone else is doing it,” “these changes are inevitable, therefore we should participate in making the changes” etc. Insofar as adoption of AI in philosophy amounts to a significant change in the status quo, it seems to me the burden of proof rests with those who advocate for adoption, and it is their arguments that deserve most of our scrutiny.
From my perspective, even as a user and defender of AI systems, this looks like a good default policy for students in coursework. I could imagine individual instructors eventually coming up with assignments for which modifications would make sense, but those would need to be carefully thought out.
I also like that this is explicit that it’s only for assessed work – there are lots of productive uses students could have of these sorts of AI systems to have additional conversations outside of coursework to help deepen and test their understanding of material (the way you can when you have access to a person who has already read all the material and more – which students don’t generally have access to outside of class hours).
The only quibble I have is with including justifications 6-9. I don’t think that the claimed justifications 6-9 make much sense as a justification for this sort of policy, any more than they would make sense of a policy of requiring students not to take the bus, because the bus uses enormous amounts of energy and is operated by profit-driven corporations (in the UK anyway – in the US, buses tend to be run by municipal governments). In any case, justifications 6-9 are external to the point of philosophy classes.
Still, if including those justifications makes the policy more popular with students, that’s a minor quibble – justification 1 is clearly the most important, and justifications 2-5 are all also important.
I’d say justifications 6-9 are both relevant and compelling to many. Even if there’s disagreement on the extent of each issue, it still seems there’s at least prima facie reason to be concerned about those things. These could be major problems, and erring on the side of caution is prudent.
But more importantly, being “external to the point of philosophy classes” doesn’t seem to be enough to reject those justifications. After all, the case for allowing students to use AI is nearly entirely about their job prospects after graduation, and that’s clearly external to philosophy classes.
Or if you think it’s fair to be concerned about a student’s future (job prospects), it seems that justifications 6-9 are also rooted in the same concern. And it’s very reasonable, if not expected, for teachers to be concerned about students’ future…
I admire this policy but the bullet below is going to face a structural challenge. It’s nearly impossible to use any modern search engine right now without getting an unprompted AI answer for your search query.
I also get unprompted email summaries in Microsoft Outlook (which can be turned off), and offers for AI assistance in my PDF reader (which I have turned off).
It’s getting to the point that I will need to stay offline completely to avoid LLMs.
I don’t think I see what the problem’s meant to be here? Being ambiently exposed to some AI junk in the course of operating my computer just doesn’t seem like the sort of thing that counts as ‘use of AI … when completing assessed work in philosophy’ (although I grant that we can probably cook up some weird cases on this). It’s true that the policy counts the cases you point to as cases of “AI”—and it seems right to me that it should do so—but I’m unclear on why that’s a bad thing.
Does the policy include preparatory activities like consulting an AI for advice on how to write the essay, brainstorm etc., even if the actual writing is being done by the student themselves?
It should.
I think it does, since such uses still participate in global capitalism.
This policy’s justifications seem poorly written and intellectually lazy, let me be blunt. This is just one case among many in which I feel like AI opponents (so to speak) get a pass for bad writing/arguments, just because their conclusion(s) sounds plausible and morally right.
Without more specific criticism, it’s hard to tell why they’re supposed to be poorly written and intellectually lazy. Just saying so isn’t any more than name calling (which one might criticize as bad argumentation and intellectually lazy)
I don’t think we yet have a formal departmental policy on AI based cheating, but this is what I tell my 200-300-level students.
DO YOU NOT USE ChaTGPT OR ANY SIMILAR SYSTEMS TO WRITE YOUR ESSAYS OR IN PREPARING YOU PRESENTATIONS
You should write your essays and write them in your own words. No ChatGPT, nor any of its competitors, including Gemini, Claude, Llama, Ernie, and Grok. Not even Grammarly. So far as I am concerned, AI generated essays are instances of plagiarism and that includes essays that are partially generated by an AI. If detected they will be treated a such (that is as instances of Level 3 Misconduct) See below for our plagiarism policy. However, you will probably be doing yourself a disservice if you submit an AI-generated essay even if I don’t manage to catch you out. For a start, you won’t have done the brain-stimulating work that you would have done if you had written the essay yourself. Essays are both a summative and a formative mode of assessment. You get a mark – that’s the summative part – but you also learn by writing them – that’s the formative part. If you don’t write your own essays you may get the summative mark – though of course it will be fraudulent – but without the formative experience. There is also some evidence that undue reliance on AIs can be bad for your brain. The Flynn effect (Look it up!) may be going into reverse.
Furthermore, cheaters are unlikely to prosper when it comes to the final grade, since they are likely to come a cropper with respect to the final exam, worth a potential 50% of your total mark. This will be three hours long with three questions to answer. AI cheats who have not written their essays are likely to to make a hash of things when answering the corresponding exam questions. You may think that it is possible to get around this by memorising an AI-generated essay, corresponding to an essay prompt, and then regurgitating it in the exam. Not so. Though there will be exam questions corresponding to most of the essay prompts, they will be questions with a twist. They will ask you to come at the topics from a slightly different angle (so to speak) thus requiring a modicum of intellectual agility and a level of understanding that transcends rote-learning. You will have to answer the specific questions that I ask, not the similar questions that we considered in class. This means that if you have not thought through the issues for yourself – specifically, by writing your own answers to (two of) the class questions – you are likely to regurgitate exam answers that respond to the questions that you hoped I would ask rather than the questions that I shall actually ask. And since I will be marking hard for relevance, there is a good chance that you will fail.
———————-
Most of my colleagues have instituted or refurbished similar assessment regimes designed to minimise the benefits of AI-based cheating.
What’s the rationale?
1) As noted on earlier threads LLM’s can facilitate the further education of the already educated. (The ‘already educated’ being those who already know ‘enough’ about a subject to learn more with relative ease.) Furthermore AIs can sometimes facilitate research. We have had plenty of testimonials to that effect.
2) However, LLM’s pose a major threat to higher education and to education generally since they make it very easy to fake understanding and to fake mastery of some skill, subject or discipline. Unless steps are taken to prevent (or mitigate) AI-based cheating, students can wind up without ENOUGH knowledge in their heads either to deserve their credentials or to use AIs (and other resources) intelligently to extend their knowledge.
The ‘enough’ is important. Obviously you can have mastered some subject even though a lot of the relevant information is NOT stored in your own brain but in books and online resources. But a substantial dollop of knowledge must be there in your head or you won’t have really mastered the subject. Suppose that economic necessity compelled me to put on a course on ‘Stuart Britain’ for the benefit of High School students, a U3A group or even a 100-level history class. I could do a decent job with a week or two’s preparation because fifty years ago we did Stuart Britain as a GCE subject and I have retained an interest in the subject ever since. Consequently a lot of information about Stuart Britain is indeed stored in my brain, and this knowledge is structured. This means a) that I would be in a position to evaluate texts and online resources and b) that having a structured understanding of the period I would be able fill in any gaps in my knowledge with relative ease. I would also know what I needed to know but didn’t. (In this area there are, for me, many known unknowns but not many unknown unknowns.) I would probably be in a good position to detect AI ‘hallucinations’. Because I have internalised ENOUGH information about Stuart Britain I would soon be in a position to lead discussions and answer questions off-the-cuff. Now suppose I was asked to put on a similar course on high-school physics. Could I do this? Perhaps (because I am a competent teacher) but not without months of preparation. The O’level physics that I did all those years ago means that I would (just about) know where to begin but I would have to put myself through an intensive course of study and it would be much harder me to evaluate texts and resources and to distinguish good information from bad. Furthermore, to begin with, I would not know much about what more I needed to know. LLMs alone would not enable me to do a decent job. The reason is that when I was a teenager I did not take as much pains to internalise what I was taught about basic physics as I did to internalise what I was taught about Stuart Britain. And in the interim a lot of that knowledge has gone to rust.
The point generalises. The things you ‘know’ as an extended mind depend upon your knowing ENOUGH as an unextended mind. In order to be a competent doctor, lawyer or engineer, you have to have internalised enough structured information to use external resources effectively, and that includes AIs. And unless the education system takes serious steps to prevent or mitigate AI-based cheating, many graduates will end up not knowing ENOUGH about anything.Furthermore they will not have learned to learn or and are unlikely to have developed the skills and capacities that we profess to teach.
Why not? Because people tend to acquire knowledge, a love of learning and the associated skills and capacities, if they are incentivised to do so. (Though we should bear in mind that some of the incentives can be, so to speak, internal, a) because learning things, acquiring skills and doing something difficult can be fun, and b) because many of our students are to some degree morally motivated and are therefore inclined to despise AI-based cheating).) However, LLMs create an environment where students are incentivised not to learn but to fake learning (since by doing so they tend to get higher grades). The robots help create an environment in which cheaters tend to prosper and in which, accordingly, many people cheat.
Thus to re-incentivise learning we have to disincentivise cheating by reconfiguring our assessment regimes, so that cheaters tend NOT to prosper. This can be done, for instance by putting a large proportion of the mark in a cheat-proof final exam. Is it otherwise undesirable to do this? Yes, because final exams are, in the odious jargon which we have to put up with, ‘summative’ rather than ‘formative’: you don’t learn by doing them but only manifest the learning that you have already acquired. In the absence of cheat-facilitating LLMs an assessment regime based on long-form take-home essays would definitely be preferable (at least in philosophy) since students learn about the topic in the course of writing their essays. So in minimising the share of the grade that goes on take-home essays (for example) and maximising the share that goes to cheat-proof mechanisms such as final exams, there is undoubtedly some loss. Thus the LLMs have done us all a damage by forcing us to revert to to assessment and teaching regimes which are otherwise suboptimal. That does not mean that we don’t have to do it.
.
A final point. I think that the graduates of universities which make a point of disincentivising AI-based cheating – for instance by making a song and dance about the policy that at least 50% of the grade in every course must be due to cheat-proof methods of assessment – are likely to be at a premium in the job market, since employers can be reasonably confident that they have earned their credentials. As for the objection that they may be slightly less adept at the use of AIs for legitimate purposes than the graduates of other universities – well, I think they will be able to make up their deficiencies in fairly short order since most AIs are designed to be user-friendly. It isn’t that hard to learn how to use AIs intelligently, especially if you know how to get by without them.
I strongly agree with the majority of this. I think where I get worried about blanket prohibitions on using LLMs is that they can obviously be used to do a bunch of things we used to encourage students to do–and which I was encouraged to do as a student. We generally want students to discuss the readings with their classmates and think it is fine to work out the correct interpretation collaboratively. We encourage them to get feedback from others on their drafts and to revise them in light of that. Universities even have writing centers that provide that feedback, and we encourage students to go to them. I would generally not have a problem with a student reading a simplified summary of some philosophical argument before reading the assigned text to help them understand the assigned text. This can admittedly create gray areas in terms of what counts as the student’s own work. Sometimes students who were discussing ideas together hand in papers in which the content is too similar, and this creates problems. But, in general, everyone up until now has seemed to think this is good, to be encouraged, and not a form of academic dishonesty.
Anyway, the problem is, students can use LLMs to do these sort of things as well. And these rules are being written in such a way that they seem to say that using LLMs for any of this is also academic dishonesty. Using them as a search engine for finding articles to read would be a form of academic dishonesty, the way these rules are written. Now maybe this isn’t intended, and we just need to state the rules more clearly. But, when I try to ask people if this is what they intend to rule out, a surprising number of them say, yes, obviously. (A surprising number also seem assume, I would add, that using Chat to search for articles means simply copying a made-up citation list that Chat has hallucinated. And yes, some students do that. But one can also use it as a very effective search engine for articles that one then reads. And students can include phony citations without using an LLM.) So I don’t think I’m just being uncharitable or nit-picky.
The other thing I have to strongly disagree with is the claim that AI software is user-friendly. It is maybe the most user-unfriendly consumer-facing software in decades. And this is why people who really use it well build harnesses and learn about hooks and a bunch of other stuff beyond my pay-grade, and why AI companies provide long PDFs about writing prompts that start with sentences like “Prompting is more of an art than a science.” The times I have gotten it to work well–aside from basic tasks like web searches–required several hour conversations with it beforehand on how to structure my instructions to get the results I was looking for, along with considerable trial and error.
I think we can all agree that tracing paper has only limited use in improving drawing skills, but it’s both silly–and detrimental to a fuller integration of opportunities for learning and problem solving–to deny tracing paper.
The parallels are not difficult to appreciate.
Baudelaire was extremely persuasive regarding the perceived harms of photography, but do we think he was right in retrospect?
Perhaps “persuasive writing” is overrated. Who wants to be persuaded of falsity? We have all wasted significant portions of our lives, perhaps more than we realise, reading persuasive writing, being carried along with the flow and seduced by the sophistication. Perhaps original thought will become more lauded in the age of AI and persuasion will be seen as a sham. That seems like a genuine gain to me.
Where sophistication and originality are really needed is in our ways of dealing with the problems AI raises. Injunctions are perhaps a useful stopgap measure but they’re not a solution that promises to really help help staff and students in the longer term.
As far as I’m aware, drawing is not typically taught through tracing, and never has been. In fact, if you enroll in a fine arts program, you will primarily be learning the trade by drawing nude models from life, under supervision.
Thanks Michel, Perhaps i should have mentioned that I’m an art teacher at not one but 2 art schools and have been for many years. We learn drawing both by copying and by observation. I didn’t say we teach the use of tracing paper but tracing paper is a great way to analyse and learn how to apply perspective. We teach gridding off images which is effectively the same and is also regarded as a shortcut by some.