Does Your Department Have an AI Policy? Hereâs Edinburghâs
Does Your Department Have an AI Policy? Hereâs Edinburghâs
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.
Itâs the default policy, which means it can be superseded by policies developed by individual instructors for their courses or approved disability accommodations.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 universityâs 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 studentsâ 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 âno AI useâ policy for students:
Justification of policy:
-
- Educational significance: The articulation and organization of oneâs 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 wouldnât 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 studentâs learning in a course. The use of AI violates the principle of academic integrity since it represents something as oneâs 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: âcognitive offloadingâ, âmetacognitive lazinessâ 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 peopleâs 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 oneâs 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 â they wanted to insist it keep being done with an axe. Guess who went out of business (it wasnât 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 didnât cause cognitive decay in the youth, although i suppose they did probably cut some people. Anyway though, your general point does seem right â which is that policies that involve hermetic seals like Edinburghâs constitute a form of denying the age we are in. so i think youâre right that if we live in an AI age, banning AI in oneâs 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 wasnât a real person donât you? I guess I should be happy you didnât 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 weâre in the age of the internal combustion engine. We have a choice about what sort of future we want. We donât have to give in to the tech bros and condemn our children and everyone elseâs to unending AI slop and brain rot.
Not to mention, we donât have to (and shouldnât) 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 Bunyanâs 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 isnât named Ichabod Crane, heâs actually a woman and he runs the local BP station instead of teaching school. And it turns out the headless horseman isnât 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.
Taken out of context, this might be my favorite Daily Nous comment of all time.
âThe future is X so we must do X is an ambitiously awful sort of argument.â What Sam Duncan said. In capital letters.
Iâd include Captain James T Kirk in the Mix: AI Fails in Star Trek: The Original Series â Practicum Strategy
âWe are in the AI age now, like it or not.â
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 canât 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 youâre advancing, though you havenât 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.
I canât believe how many people are still advancing this specious argument. Can it really be the case that so many of us forget the simple fact that we are only a small part of our studentsâ education? No philosophy department has the power to hermetically seal a student off from anything. That ought to be obvious, and so it ought to be equally obvious that even a very strong blanket ban on AI use in philosophy courses will have only a very small effect overall on a studentâs AI use in the rest of their education, to say nothing of the rest of their life.
No one who grows up in this ridiculous world we have created will be able to avoid being dragged out into this particular âsunlightâ (irony of ironies!). We philosophers might in the meantime actually think about what we want students to learn, and be honest about the dangers AI poses to that. Bravo to Edinburgh for at least taking a stand.
This is not a very good line of argument, Bilingual. The underlying idea of the DLC post doesnât seem to be âwe are in the age of X, therefore we should just succumb to Xâ (thatâs a strawperson) but rather, and more charitably to DLC, âWe are in the age of X, so we should have policies that appropriately acknowledge thisâ, and thus, when those policies are pedagogical policies, theyâd take the shape of acknowledging that the world (where jobs are to be found, and skills we learn in education are to be deployed) is an AI integrated world. In your nuclear bomb example, acknowledging being in such an age (in line with the general DLC principle) would look like (pedagogically) preparing students to go into a world that includes nuclear bombs; not acknowledging being in the nuclear age pedadogically speaking would be to educate students in such a way that theyâd be preparing for a non-nuclear world. That is obviously manifestly unhelpful, which I take it is what DLC is saying about the Edinburgh policy.
Simon, I think you are getting this analogy wrong. Bilingual and Patrick Lin, etc, seem to be suggesting that being in an age of AI is something we should all âacknowledgeâ but then resist, in the way that we can acknowledge pollution and toxic chemicals in an environment but then wish to occupy, say, a part of a building or a room that is not contaminated by it. Wanting the chemicals not to touch you is not a failure to acknowledge, but more like a tacit adoption of a rule that says âif you identify X somewhere locally, then avoid X.â That at any rate is the proximal rule, where the more distal rule we can attribute to their position (it seems any way) is something like âif X is globally in an environment, then both attempt to resist/avoid X locally, while taking actions to make it such that X is not locally/globally presentâ, and where the latter distal principle is a more full blooded crusade, as opposed to the kind of local protectionism from the proximal principle.
âAI existsâ 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 wonât make you strong. Itâs not denialism about the machine age that at play here: if youâre stocking a warehouse, the forklift is a great tool. But if youâre trying to get strong, the forklift is stupid â more precisely, youâre 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 youâll trick even more people into responding to you.
Compare these accusations of âdenying the futureâ with the fact that, âwhen 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, donât worry: Delineating a space for trees in a time of rapid urban expansion wasnât âbackward-looking.â On the contrary, it was VERY forward-lookingâ (taken from https://certifiedaifreeskillsandknowledge.org/).
Thatâs a beautiful analogy but people still want to live in their apartments, not in the park.
Good. This shows that â[m]aking your courses AI-free wonât âdepriveâ 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 climbâ (pasting again from https://certifiedaifreeskillsandknowledge.org/).
I donât 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?
The difference is that someone advocating for the use of chainsaws at a university convocation would not have gotten booâd offstage.
History is not inevitable. Nor should it be decided by a bunch of oligarchs who have no concern for economic equity or the environment. Bravo PHIL at EDI!
So reasons 3, 7 and 9 would all seem to pretty straightforwardly rule out using Microsoftâs grammar checker.
Agree â 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 donât 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 thatâs 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.
Whatâs different with grammar checkers is that some, notably Grammarly, have now evolved to be much more than mere grammar checkers. So, itâs 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 didnât miss anything. Thatâs exactly what I was referring to. Grammar checkers definitely violate 7 and 9. And I canât think of any reason why using an LLM to check grammar would be ruled out by 3 that isnât 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 didnât 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?
Youâre 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 weâve chosen to grandfather them in because everyone is so used to them?
One could also use an LLM as a simple word processor. And you would probably be right that using it that way would not technically be ruled out by any of the rules. But surely it is clear why you probably wouldnât want to make that exception. Any allowed use of an LLM makes illicit use one step easier.
First of all, if thatâs the rationale for a total ban, say so.
Second, how would it become easier? Itâs incredibly easy to do either way. The difference in difficulty between opening up one of these platforms to cheat and cheating when already at the platform is trivial.
You are right that there is barely any difference physically. But I would argue that there is more of a mental barrier when it is banned completely. Once you allow something a little bit it is easier to convince yourself that it is ok to use it a little bit more.
I think a lot depends on how the person in question is thinking about the activity. Allowing myself to drive doesnât make it easier for me, I think, to allow myself to run people over. Iâm not sure that allowing oneself to use AI as a proofreader makes it easier to use it to cheat. But if it does, Iâm not really sure that banning its use on *assignments* does much, since students will allow themselves to use it for other things, and make it easier to allow themselves to cheat that way.
Iâd argue that allowing yourself to drive and allowing yourself to run people over are far enough apart that the analogy doesnât work. Allowing yourself to kill (or at least maim) people is drastically different from allowing yourself to drive. I would equate it more to cheating on a diet. It is much easier to convince myself that a few cookies are fine once I have already âbroken the sealâ and had one cookie.
Iâd also argue that allowing even a small amount of AI use makes it easier for students to come up with an excuse if they get caught going over the line slightly. Imagine that a student is using AI to do just a bit more than proofreading. Perhaps they are letting it re-write full sentences. It seems plausible that they might think they can get away with claiming that they just didnât understand where the line was. But if the line is very stark it becomes much more difficult to make a claim like that.
How are we defining âfar apartâ and âcloseâ? Yes, obviously driving a car and running people over *feel* far apart to most people. But you are making claims about what will feel far apart and close to a student. Maybe?
Iâm sure students who get caught cheating will have all sorts of excuses. So what?
They explicitly state that they only allow grammar checkers âthat conform to the universityâs Proofreading Policyâ. To me this seems like a reasonable sort of policy â âgrammar checkersâ 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 donât think reasons 6-9 are great reasons for the policy.)
Theyâre inconsistent with the policy.
can you state how you see this to be the case?
âStandard spell-checking software (e.g. in Microsoft Word), citation management software, and grammar checkers that conform to the universityâs Proofreading Policy are excluded from the definition of AI.â
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 âarticulation and organization of thought.â while grammar surely plays
a role in articulation and organization of thought, itâs 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 donât 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]), iâm not sure i see your case.
Iâve 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 thatâs the principle thatâs 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 donât really understand what that has to do with the issues Iâm noting with these principles.
Sorry about the double-reply. The first one didnât show up on my end until several hours later.
Hear hear!
This is one of the best AI policies Iâve seen in education. Finally!
Itâs clear, brief, but also detailed enough to be substantial, persuasive, and a nice starting point for discussions.
And itâs flexible enough (in that it may be superseded by an individual instructorâs 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!
đ
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 donât 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. Iâm on the fence about 5 because my impression is that the jury is still out on âcognitive harmâ (a loaded term) and much of what is established is already covered by 1 and 3, so itâs not doing any independent work other than using scary words on the basis of contested evidence. BUT Iâm willing to grant that extensive use of AI is bad for studentsâ intellectual development in some sufficient sense. I agree about 6-9 and enforcement.
Sorry I meant 1-3 and 5 and Iâm on the fence on 4.
I think this underestimates the difficulty. Enforcing such a rule isnât as simple as noticing the clear AI tells and giving the student an F. Thatâs 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 â at least, not to the satisfaction of a dean or a provost whose eye is on the universityâs 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. Iâve been able to catch students without ever relying on such tools, but only because the students confessed.
Exactly. And it will get harder to get students to confess, because they will come to realize that if they donât, thereâs ultimately nothing we can do.
One more reason to drop justifications 6-9! Justifications 1-5 are sufficient to justify students not using AI for a lot of school assignments, but they provide no reason not to use AI tools in application outside of study (i.e., when trying to catch violations).
Given the energy consumption of the gigantic data centers required, how could one seriously deny environmental costs?
All costs are relative to the alternatives, but they also tend to be overstated. Andy Masley has been doing the lordâs work on his Substack to debunk some of that misinformation.
Iâm very curious about this, as this issue is important to me. Iâll try to review some of Masleyâs data on my own time, but could you expand on what the alternative is, in this context? It seems to me that the alternative would just be to refrain from building huge data centers that gobble water and power, but Iâm not as informed on this as I should be. Are there any positive effects of AI on the environment that offset the cost of resource consumption? (sincere, not rhetorical, question)
Iâm talking about the alternatives to AI use, not to mention the myriad activities you tolerate that also require data centers, carbon emissions, and so on. For any task you want to complete, there are several ways you could complete it. LLMs have a non-zero environmental footprint, but so does almost everything you do and consume, so you have to compare the costs of your executing tasks using LLMs vs the costs of alternate means to completing your tasks. Granted, if everything you could do with an LLM you do manually, or relying on old school methods, eschewing the web as much as possible, you are likely to have a lower footprint. But most of the digital tools you use have a non-zero footprint (especially online videos and Zoom), so the question is how big the delta is. But even bracketing comparisons for a moment, if environmental concerns were a sufficient reason to eschew AI use, they would require you to abstain from many practices that a philosophy department just has no business policing, from how you commute to what you eat, from how you thermoregulate your house to when and where you travel. Lots of good reasons to try and minimize your impact there but Iâm surprised how many philosophers buy the environments objections to AI who otherwise donât mind sharing reels on Facebook or Instagram, spend hours everyday on Twitter or Bluesky, and consume more meat and dairy than most people ever have in the entire history of humankind.
But yes, you should read Masley.
Fair point. And I agree that none of us are free of hypocrisy, in this regard. Obviously, if we were to align our actions perfectly with our principled opposition to environmentally deleterious activities, we would have to make dramatic changes (for what itâs worth, many of us do actively try to do our best, in this regard). However, your comment seems mainly to emphasize the hypocrisy/inconsistency issue, without presenting any examples of instances in which AI is actually better for the environment than alternatives that donât involve AI. Are there any? If not, then it seems like your point is simply that we would have to inter alia âeschew the web as much as possible,â if we were to be consistent. I agree, but thatâs not really what Iâm wondering about.
I didnât claim AI was better. I only implied that the difference might not be large enough to be that concerned. But Iâd wager, for instance, than an hour on YouTube or Zoom is typically a lot worse than an hour on Claude. Itâs also possible, though Iâm not as confident and I donât think weâre there yet, that efficiency gains afforded by AI will someday be large enough to offset their additional costs compared to alternatives. Whatever is the case, the environmental objection is typically based on mistaken information and formulated in a vacuum. And more importantly, just not the sort of thing a department should regulate.
With all that said, there is no denying that new data centers are incredibly energy intensive, though this long predates LLM adoption.
Understood. Thanks for taking the time to explain
You bet. And thanks for being patient with me!
Itâs not your main point, but I just want to flag that the idea of âeschewing the web as much as possibleâ for environmental reasons most likely gets things backwards! Driving to the library, for example, is going to do orders of magnitude more environmental harm than a day of typical consumer web use (including LLM use).
As Masley observes: âThe internet is ridiculously energy efficient compared to most other ways we spend our time. It takes up 50% of our time, but is only 12% of how we use electricity.â (Or just 4% of overall energy use.)
Digital tools are vastly more efficient than physical ones. It would take something like 10,000 chatbot prompts to rival the energy costs involved in producing a single physical book (mostly due to paper manufacturing). And thatâs not even counting the Amazon delivery.
As Masley mentions in his overview post, you likely waste more water by walking outside (due to the slight wear on your sneakers, which require a lot of water to replace) than by sitting at your computer and prompting LLMs non-stop.
In short: computer use appears to be among the most energy-efficient, environmentally friendly things we spend our time on. Almost everything else we do is worse.
(âData centers are energy intensiveâ is like âCities are energy intensiveâ. True but misleading. Cities concentrate population in a way thatâs much more energy-efficient than having a similar population dispersed. Data centers do a similar thing for computation.)
Oops, mangled my second sentence there. It should be something like âorders of magnitude more environmental harm than researching using the web (including LLM use)â, or else that it may be somewhere in the vicinity of a full day of consumer web use. Apologies for the mixed-up exaggeration!
Great point, I agree. But I imagine one could just stay home and use oneâs personal library!
You have to amortize the costs over the number of users. Even if we donât know the total emissions caused by these huge data centers, we do know that the emissions result from the companies spending money on electricity and manufacturing. Furthermore, we know that money spent on electricity and manufacturing generates no more emissions than spending that same amount of money on gasoline to burn. Finally, thereâs good reason to believe that the companies are actually breaking even on revenues vs inference costs, with training costs being at most a factor of 2 or 3 beyond these inference costs.
Putting this all together, we can be pretty confident that a user on a $20/month subscription is causing less emissions than someone who spends $20/month on gasoline â and probably a lot less. (And in terms of water use, even the high estimates show far less water use than spending $20/month on meat or dairy.)
Thatâs not to say itâs a non-issue â but just that itâs an issue on the same order of magnitude as faculty commuting to campus to do their work, or failing to keep a vegan diet.
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 arenât always made explicit, are not exactly airtight: âeveryone else is doing it,â âthese changes are inevitable, therefore we should participate in making the changesâ 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 itâs only for assessed work â 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 â which students donât generally have access to outside of class hours).
The only quibble I have is with including justifications 6-9. I donât 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 â 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, thatâs a minor quibble â justification 1 is clearly the most important, and justifications 2-5 are all also important.
Iâd say justifications 6-9 are both relevant and compelling to many. Even if thereâs disagreement on the extent of each issue, it still seems thereâs 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 âexternal to the point of philosophy classesâ doesnât 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 thatâs clearly external to philosophy classes.
Or if you think itâs fair to be concerned about a studentâs future (job prospects), it seems that justifications 6-9 are also rooted in the same concern. And itâs very reasonable, if not expected, for teachers to be concerned about studentsâ futureâŠ
I just donât see how justifications 6-9 give any more reason to regulate student use of AI than they give to regulate student consumption of meat, student use of buses and other fossil-fuel-powered vehicles, or student wear of synthetic fabrics.
I think you mean only justifications 8-9 can also apply to regulating meat consumption, etc.?
That may be true, but we donât need reasons bespoke to AI, do we? Many of the other justifications (e.g., 1-4) can also apply to straightforward plagiarism, e.g., copying off Wikipedia, etc., but you donât seem to have a problem with those reasons.
Anyway, I still donât see that the internal-external distinction is relevant to how appropriate a justification isâŠespecially since the case for allowing AI use relies nearly entirely on external reasons (getting a job)âŠ
I admire this policy but the bullet below is going to face a structural challenge. Itâs 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).
Itâs getting to the point that I will need to stay offline completely to avoid LLMs.
I donât think I see what the problemâs meant to be here? Being ambiently exposed to some AI junk in the course of operating my computer just doesnât seem like the sort of thing that counts as âuse of AI ⊠when completing assessed work in philosophyâ (although I grant that we can probably cook up some weird cases on this). Itâs true that the policy counts the cases you point to as cases of âAIââand it seems right to me that it should do soâbut Iâm unclear on why thatâs 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 policyâs 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, itâs hard to tell why theyâre supposed to be poorly written and intellectually lazy. Just saying so isnât any more than name calling (which one might criticize as bad argumentation and intellectually lazy)
Iâd admit very willingly that I am lazy when posting a comment here. But I would be much more strict on myself when writing an official policy. My comment would be much better if I were less lazy, but this is a tradeoff!
AI opponents get a pass compared to whom? AI boosters? Please be serious.
I donât 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 donât manage to catch you out. For a start, you wonât 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 â thatâs the summative part â but you also learn by writing them â thatâs the formative part. If you donât write your own essays you may get the summative mark â though of course it will be fraudulent â 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 â specifically, by writing your own answers to (two of) the class questions â 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.
âââââââ-
Most of my colleagues have instituted or refurbished similar assessment regimes designed to minimise the benefits of AI-based cheating.
Whatâs the rationale?
1) As noted on earlier threads LLMâs can facilitate the further education of the already educated. (The âalready educatedâ being those who already know âenoughâ 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, LLMâs 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 âenoughâ 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 wonât have really mastered the subject. Suppose that economic necessity compelled me to put on a course on âStuart Britainâ 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 twoâs 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 didnât. (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 âhallucinationsâ. 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 Oâlevel 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 âknowâ 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, âsummativeâ rather than âformativeâ: you donât 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 donât have to do it.
.
A final point. I think that the graduates of universities which make a point of disincentivising AI-based cheating â 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 â 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 â 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 isnât 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 doâand 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 studentâs 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 isnât 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 donât think Iâm 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 âPrompting is more of an art than a science.â The times I have gotten it to work wellâaside from basic tasks like web searchesârequired 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 itâs both sillyâand detrimental to a fuller integration of opportunities for learning and problem solvingâto 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 âpersuasive writingâ 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 theyâre not a solution that promises to really help help staff and students in the longer term.
As far as Iâm 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 Iâm 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 didnât 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.
Yes, but I think the disanalogy here is both significant and relevant. The copying that occurs in fine arts classes (what used to be called ârhetorical imitationâ) takes place in a context where (1) students are explicitly instructed to do it, as a pedagogical technique, and (2) everybody knows thatâs what theyâre doing; thereâs no attempt to pass that work off as a student original (for those who donât know, when students copy paintings, they are explicitly instructed to alter the painting in clear ways, typically by resizing it, so that it cannot be mistaken for a forgery). Thereâs no intent to deceive.
What students, left unsupervised, are doing overwhelmingly is using chatbots to produce straight-up forgeries. The pedagogical value of that is pretty much nil.
Edinburghâs policy leaves room for instructors who wish to experiment with a kind of rhetorical imitation using chatbots. But absent such a countervailing instruction, it makes the departmentâs expectations of students very clear to everyone involved.
As we all know, Charles Baudelaire is famous for his common-sense judgements which anticipate our current understanding of the world. Also, all technologies are the same. Why policing nuclear bombs when humanity has not killed itself by knives?
Do you think it would be reasonable for an art department to have a policy that says âby default, all classes will require all assignments to be completed without the use of tracing paper â however, a few classes may have other policies that do involve the use of tracing paperâ?
Jim, I donât think the tracing paper analogy quite works. From an educational point of view, the skills needed to generate an AI image or an AI essay are the skills associated with the student refining their their ability to develope prompts for gen AI. Tracing, by contrast, exercises a different set of muscles â both metaphorical and physical muscles, because tracing does in fact build up the literal muscles students can then use for free drawing. I feel the art equivalent of using AI to write essays would be students using gen AI to refine their prompting skills to create images that they can include in their portfolio, which given the changing job market is probably a worthwhile skill to learn. In short, while I can see the a case to be made for teaching students AI prompting skills, I donât feel that AI prompting skills are analogous to using tracing paper when teaching art.
Iâm surprised that people are overjoyed by this policy, but unfortunately, itâs not really surprising either. The policy does not seriously reckon with how many disabled students use AI not as a replacement for thought, but as a complex mode of performance in which AI assists without supplanting their philosophical work. A student with dyslexia, a linguistic processing difference, or a motor disability affecting writing may use generative or paraphrasing tools to externalize thinking they have already done, getting ideas out of their head and onto the page in a form that communicates their reasoning. Collapsing this use into âacademic misconductâ unless a formal accommodation is secured places the burden on disabled students to medicalize their practice in order to access tools that non-disabled peers simply donât need.
What is indefensible is the slogan âEdinburgh philosophy is human-made!â implying disabled students are not fully human, or worse, that they are the presumed victims of âcognitive offloadingâ and âmetacognitive lazinessâ simply for using tools their neurology requires. This is concerning, to say the least.
The justifications under âEducational Significanceâ and âIntellectual Autonomyâ rest on a normative picture of how philosophical cognition properly proceeds: thought is organized internally, expressed linguistically through oneâs own unmediated effort, and that process is essentially the practice, the raison dâĂȘtre, of philosophy. This is neurotypical and is not an accurate picture of what it looks like to think for oneself. It assumes a single cognitive architecture as the standard against which deviation must be excused. Philosophers working in the philosophy of disability have argued that this kind of universalism smuggles in particular embodied norms under the guise of describing cognition in general.
The accommodation exception doesnât resolve this. It treats unassisted neurotypical cognition as the default and grants exceptions at the margin. A more philosophically honest and interesting policy would interrogate that default rather than institutionalize it.
Right, and the institution may find itself being sued: https://www.independent.co.uk/news/world/americas/university-of-michigan-ai-cheating-lawsuit-b2921621.html
Iâm quite sympathetic to disability issues, but are you saying that we shouldnât treat the neurotypical asâŠthe typical?
Because thatâs what the AI policy seems to be aimed at. It wasnât aimed at everyone (given the accommodation exception, etc.) but only to the lowest common denominator or the broadest, most typical audience.
If you donât like the word âdefaultâ, ok, thatâs fair. But I donât think the policy is dishonest just because it doesnât offer a unified theory of responsible use in AI that speaks to all demographics, or n number of theories for n number of demographics.
That said, itâd be interesting to hear if you have a better way that solves the concerns you mentioned.
If their disability really requires them to use AI (which I doubt), they can get an accommodation.
How many disabled students use AI not as a replacement for thought, but as a complex mode of performance in which AI assists without supplanting their philosophical work?
If weâre going to reckon with that, letâs reckon with it.
#9 is embarrassing, but more importantly, it is wildly inappropriate. This proclivity to abuse power to impose and conscript others in oneâs politics is so quintessentiallyâŠwoke (in the worst sense). Boundaries! Not since Rick James have we seen such habitual line-steppers.
You donât see that this one little university is trying to combat a much, much larger abuse of power by AI companies, which have a combined market cap in the trillions of dollars?
Or if youâd say that those companies arenât abusing power but simply acting their part in a market economy, then so is the university or anyone else who wants to warn people away from a product or service.
Anyway, no, thereâs no abuse of power here by Edinburgh. Itâs just a baseline policy that can be easily superseded, i.e., itâs not even binding. And even if this were an abuse of power, itâd be justified and reasonable as a counterbalance. Unity!
Hi Patrick,
I disagree.
A university departmentâs proper mandate does NOT include engaging in this kind of political combat, whether the target is AI companies, unions, foreign governments, abortion clinics, Blacks Lives Matter, or Wall Street. Itâs not about the merits of the issue; itâs about the role of university departments. The individuals who comprise the department are welcome to engage in all the political combat theyâd like in their capacities as private individuals without leveraging their control of a university department. Including #9 in their policy justifications violates this principle. In this case, the abuse of power boils down to an improper, superfluous justification of a policy (for which 1-3 is sufficient); so, itâs not concrete. But the principle still matters, both in itself and because (adding to) the precedent of violating it has implications elsewhere that are concrete.
Thatâs not unity; itâs annexing the department in the service of oneâs politics.
Are you denying the âmuch, much larger abuse of power by AI companies, which have a combined market cap in the trillions of dollarsââor are you just ignoring it and think itâs more important to tell a university department what their âproper mandateâ should be?
This frame and the approach you are taking are incompatible with a liberal and pluralist academy.
I hope you donât mind me jumping in here.
You say, âIncluding #9 in their policy justifications violates this principle [of standing back from political combat when in oneâs official university role]â
Accepting that there is some such principle, I think youâre not giving due consideration to what counts as political in the verboten sense and what doesnât. Engaging in partisan politics is one thing; pointing out the structural conditions â the socio-political-economic landscape in which we work (incentive structures of AI development e.g.) another.
Hi Runa,
Thanks for joining.
I would carve it up according to the different roles faculty members and departments play and the different reasons that are appropriate considerations within each role. So, for a given issue, there can be department policies and actions that are appropriate, for which some reasons are appropriate considerations, and some arenât. If appropriate reasons are sufficient for the policy, proceed; however, if the case for the policy depends upon inappropriate reasons, donât proceed. The inclusion of inappropriate reasons in the justification of a policy is inappropriate.
In Edinburghâs case, âMost AI systems are produced by profit-driven corporationsâ is an inappropriate reason for a department AI policy. If the justification of their AI policy depended upon that reason, it would be an inappropriate policy; however, since #1-3 are sufficient reasons for their AI policy, the policy is unproblematic. Nonetheless, their inclusion of #9 in their AI policyâs justification is inappropriate.
I am extremely worried about what the development of more and more advanced AI will mean for academia in general and academic philosophy in particular. However, this restrictive policy has at least three very severe downsides, namely:
.
Itâs not âcreating the strongest work possibleâ when students use it, period. This is to confuse process and product. I do not care about their 3 page essay on Aristotleâs theory of friendship; I know that stuff already. What I care about is that they learn the things one learns in the course of writing the 3 page essay on Aristotleâs theory of friendship. If they copy another studentâs paper, or copy off wikipedia, or ask ChatGPT, theyâre not learning those things. The entire purpose of assigning an essay is for _them_ to do the work â all of the work; the brainstorming, outlining, research where applicable, and yes, constructing sentences. Using the robot to do any of this defeats the purpose of assigning essays (which is why Iâve pivoted almost exclusive to blue book exams).
I think the charitable construction of the problem is that unless we want to say that discussing the topic with your friends or asking them for critical feedback about drafts or reading secondary literature is also forbidden, which it didnât used to be, there are ways of using AI that arenât the same as copying. And maybe we have to forbid those uses too because otherwise the rule is unenforceable. But the rule isnât enforceable anyway. So it ends up feeling unclear what the goal is.
But yeah, Iâm using blue book exams next semester as well.
I canât speak for Simon. But I do think this ban on AI is unenforceable. What I mean is that itâs administratively unenforceable â or at least it will rapidly become so. I think it will take a university with a remarkably stiff spine to continue to support faculty against an increasing swell of students who are complaining that they have been given an F for using AI without any proof that they did so. When push comes to shove â as it will â the faculty memberâs confident say-so that the student used AI will not be enough. The dean or the appeals board will demand proof, and that will not be possible. No AI detector, no matter how low its false positive rate, will suffice when put up against a student who swears up and down that they didnât do it.
Iâm certainly willing to be proven wrong about this. But this is how I think it is going to go. Students will realize that all they need to do is deny the charge; and administrators will cave because they donât want to upset so many students (and possibly also risk lawsuits from some of them).
This is my situation. We have an academic integrity policy at my institution that says evidence from âAI detectorsâ is inconclusive and so unless the student is exceedingly poor at covering their tracks, it is basically impossible to have a charge of AI misuse stick unless the student confesses. Since theyâve already submitted noncompliant work in direct awareness of a class assessment policy that restricts its use most of them just brazenly stonewall and there is nothing that can be done.
Like others, I donât care about the product of assessment. Iâm aiming at the cognitive development and habits of thought fostered by the assessment process. Moreover I donât think it will be hard for my students to learn how to use these tools in their workplaces, if needed. (Prompting a chat bot is not like parsing Aristotle so I reckon my students can handle it.) So handwritten exams it is (with appropriate accomodations).
Oh man. Iâve given up entirely on discussion posts. Theyâre super demoralizing for the few students who donât use a chatbot, and generate a flood of complaints from the others who do.
It sounds like you have a lot students who are using dodgy âwrite my essayâ websites with old models and you think youâre catching most of them.
People have been dropping this cheap rhetorical move on me for two years now when I say Iâm not all that impressed with AI writing: Oh thereâs some super awesome AI you obviously donât know about or you would be! Yet they never bother to show me this amazing work itâs producing. They usually even tell what it is. (I mess around with them to know it ainât Claude or Chat). Itâs basically âWho are you gonna believe me or your lying eyes?â which isnât and never will be a good argument.
Itâs not a cheap rhetorical move. And youâre asking me not to believe *my* lying eyes. I honestly havenât seen the problems you are talking aboutâmassive howlers about well-known philosophical positionsâin 2.5 years. (With maybe one exception, now that I think about itâŠ) I did see them a lot 2.5 years ago. I donât anymore.
I believe you are seeing this. Iâm just telling you, Iâm not. If you are wading through a mass of papers every semester that have these particular kinds of mistakes, your students are using much worse models than mine are. This isnât even to say that more up-to-date models are amazing, just that they do consistently produce better work than that. They produce work that is good enough that I canât simply fail it on the basis of being ridiculously incompetent. If I could, these academic dishonesty policies wouldnât even be necessary, because students would quickly realize they are going to fail if they use AI.
There are still tells for AI papers, and a lot of those papers have problems serious enough that I can point to academic dishonestyâcitations are so bad in various ways that I can say the student is citing work they never read. But usually bad AI papers have a coherent and plausible overall argument. Itâs at the paragraph level that things are really sloppy. But if a student did that, it would usually be a bad but passable paper. Five years ago, I honestly would have felt proud that I was getting my weaker students to be able to do that. Now I think itâs probably a machine.
I also strongly suspect I am getting a lot of hybrid papers. That is, the student is putting in some work, but also letting the AI do a lot of the work for them. Because I see papers with very original ideas and fairly tight, controlled argument, but also occasional verbal tics that indicate AI writing. So I think they are doing some work, and providing some originality, and maybe understand the topic well-enough to tell the AI to rewrite bad passages. But they arenât doing as much work as I want them to do. Or maybe theyâve picked up on some of these verbal tics as what you are supposed to do to sound serious or academic. Itâs really hard to tell. In any case, I canât fail them for overusing a certain kind of rhetorical device, either given university policy or frankly as a matter of basic ethical decency.
When people are saying there is a giant problem with enforcement, this is what they are talking about. You donât have to believe these models are amazing. You just have to think they are good enough to create lots of hard to detect cheating undermines the goals of the assignment, and that only the most incompetent uses are provable.
Finally, someone pasted an example of Claude below. I donât think itâs amazing. But it is certainly competent.
This is always going to be the vast majority of student (or, indeed, popular) use. Indeed, that was the vast majority of cheating before chatbots could do it for you. Some students shelled out for contract cheating, which was the gold standard, but it wasnât all that many of them.
Please see below. My students are, with maybe a few exceptions, using much better models to cheat, so it is very hard to catch them. When people complain about unenforceability, this is what theyâre talking about. If your students are handing in papers with ridiculous mistakes in 2026, they are really bad at picking which AI to cheat with.
The use of software like Pangram is ruled out by the Edinburgh policy.
Iâm sort of getting the feeling people donât care about how the technology they use works.
The point of a class isnât to get students to produce the strongest possible work â itâs for the students to develop their skills to the greatest level. Having students take a test in-class closed-book with limited time prevents them from producing the strongest possible work, but it can be a useful test of some of their skills.
True. And I am very much in favour of in-class closed-book exams.
The point about strongest possible work is more in relation to theses, bachelor and master, in particular PhD. These days, a PhD thesis can be made much stronger by AI use. It seems problematic to me to prohibit this.
Couple questions: How does using AI make oneâs PhD thesis stronger? What sense of âstrongerâ? What does it mean to say youâve earned a PhD if the robot is doing your work?
Thereâs a lot of ways you can use AI to make a PhD thesis stronger. It can help you write code in programming languages you donât know, to draw diagrams that help you more effectively convey ideas that are hard to convey through text. It can help you code up a greater variety of computer simulations to demonstrate an effect. It can be a conversation partner while you read a mathematical proof, helping you understand steps the author left out as âobviousâ or understanding what goes wrong when you change some assumptions. It can give you literature reviews of topics slightly outside your specialty, to help you understand which articles from that literature you should look at more closely. It can help you transfer your disorganized website/CV/folder of downloaded .pdfs into a file system that makes it easier to manage these things going forward.
Iâm less optimistic that it can be useful as a direct conversational partner while batting around ideas in development â but it might sometimes be better than nothing for people who are the only person in their department working on a particular topic. I donât think that having an AI system write text that will actually go in the body of the dissertation is likely to be valuable.
I think itâs worth considering that a serious AI policy cannot only be a student conduct policy. I think that it also has to be an assessment policy, and beyond that an institutional policy about what kinds of teaching are practical, fair, and properly supported.
I am not suggesting that cheating becomes acceptable when cheating becomes easy. It does not. But I do think the ease and availability matter (compare theft vs online piracy). And so, as much as we should expect virtuous behavior from students, we were never really faced with these temptations. It makes sense then that the assignments themselves should be conducive to virtuous choices (in-class drafting, in-class essays, etc). That then places a further responsibility on the institution to support instructors given that non-traditional assessment might place additional burdens.
I used Claude with a set of argument analysis skills developed by Marco Meyer to evaluate the policy. I paste here verbatim the result:
Iâll analyse the policy with your argument tools, but let me pin down scope and format first.
## Analysis of the Edinburgh âNo AIâ Policy
A structural point first, because it organises everything below. The nine justifications fall into two families that do very different argumentative work:
Only the first family bears on the policyâs actual conclusion, which is a rule about assessed work in a philosophy course. The second family, even where sound, supports something else (procurement policy, divestment, regulation), not a course-assessment prohibition. Bundling them inflates the apparent weight of the case. Iâll critique each, then build the strongest version that survives.
A second thread runs through 1â4: each treats âusing AIâ as AI does the thinking and writing instead of the student (the substitution assumption). None engages the spectrum between substitution and assistance (brainstorming, generating an objection to rebut, checking a chain of reasoning). Once that distinction is drawn, the strong arguments support banning substitution â which integrity rules already cover â not the blanket scope of âwriting, revising, and editing.â
1. Educational significanceReconstructed: writing/organising oneâs thought constitutes philosophy (P1); AI use avoids that activity (P2); so it hinders learning (C).
The inference is valid only under the substitution assumption. P2 overgeneralises: not all AI use âavoids the activityâ â using AI to surface a counterexample one then evaluates is engagement, not avoidance. The gym analogy is a good intuition pump but cuts against the scope: we donât ban machines, trainers, or spotters at the gym; we ban having someone else lift for you. So the analogy licenses a ban on cognitive outsourcing, not a ban on all AI assistance. Moderate as stated; strong only for the substitution case.
2. Academic integrityThe strongest of the nine â but it proves too little for a blanket ban. Dishonesty requires misrepresentation. If a student discloses AI use, there is no passing-off and hence no integrity violation, yet the policy bans disclosed use too. So this argument supports a disclosure requirement, not prohibition. P2 (ârepresents something not oneâs ownâ) again presupposes substitution. And the âmanifestation of oneâs own learningâ principle, applied consistently, indicts permitted aids too â human proofreaders, writing centres, supervisor feedback, citation managers all mediate the final text. The carve-out is in tension with the stated principle.
3. Intellectual autonomyLargely a re-description of (1) in the register of integrity â redundant. The distinctive claim, âfailure to think for oneself,â faces a direct counterexample: a student who drafts alone, then has AI generate the strongest objection and answers it, has arguably thought more for herself. The argument also needs a principled reason why mediation by a tutor or peer counts as good pedagogy while mediation by AI counts as a failure of autonomy. None is given.
4. Cognitive harmThe empirically load-bearing argument, and the one that could ground a genuine prohibition if the evidence supported the breadth of the claim. It doesnât yet. The cited work (a Microsoft self-report survey, an SSRN preprint, a BJET study) largely shows correlational or self-reported effects in substitutive usage; it doesnât establish that scaffolded or disclosed use causes lasting harm. More deeply, âcognitive offloadingâ is not per se harmful â writing itself, books, and calculators are offloading (Platoâs Phaedrus worry). The premise needs offloading that is pedagogically damaging in a way distinctive to AI, and the studies donât deliver that at the needed generality. Conclusion that follows: a case for howAI is taught and sequenced, not for prohibition. Moderate, with an evidential-quality caveat.
5. QualityNon sequitur as a ground for a ban. If AI output is shallow or false, normal marking penalises it â that is a reason it is self-defeating for students, hence a reason the ban is less necessary, not more. It also pulls against (2) and (4): if the work is poor, the worry about students passing it off for high marks weakens. And it is a claim about current output, so it cannot stably ground a standing policy. Weak.
6. BiasA reason to read AI output critically, not a reason it undermines assessment. The assigned canon and the open web are also biased; we respond by teaching critical reading, not by banning the sources. Applied consistently, the principle proves too much. At most supports âuse criticallyâ â which is itself a philosophical exercise, mildly in tension with (3). Weak as support for prohibition.
7. PiracyTwo problems. First, âpiracyâ begs a genuinely unsettled legal/moral question (the fair-use / learning-exception debate over training data) by labelling it. Second, the inference from âtrained on unconsented textâ to âusing the product is collusionâ needs a complicity principle that isnât supplied â we donât generally treat downstream use of ethically-tainted infrastructure as collusion. The âwe contribute promptsâ point is weak (user prompts are not training data for most assessed-work tools). Thereâs also an institutional tension: the university hosts ELM tools while the policy bans them on piracy grounds. Weak, and scope-mismatched: an industry objection, not an assessment one.
8. Environmental impactSame scope mismatch, plus an equivocation between training-scale energy and the marginal footprint of a student query (small). Applied consistently, the principle would target conference travel, cloud storage, and GPU use across the university â selective application looks like special pleading. At most a reason for moderation. Weak here.
9. Political significanceRhetorically strongest, argumentatively weakest. The corporate-profit premise flirts with the genetic fallacy: the producerâs motives donât bear on whether the tool harms learning. The anti-âinevitabilityâ point is fair but only establishes that we have a choice â it does not establish that prohibition is the right choice (non sequitur). It rests on a false dilemma: prohibit, or capitulate to âcorporate takeover,â omitting the middle of critical, regulated integration. And thereâs a performative tension â the policy elsewhere prizes autonomous critical thinking, yet here recommends shielding students from the technology rather than equipping them to think critically with and about it.
Cross-cutting weaknessesProves-too-much / consistency. The permitted-tools carve-out (spell check, citation managers, human proofreaders, âgrammar checkers conforming to the Proofreading Policyâ) is conceptually unstable against the banned âautomated paraphrasingâ and Grammarly â these sit on a continuum, and the principles in (2) and (4) would indict the permitted side too.
Strengthâforce mismatch. The justifications assert categorical wrongs (dishonesty, piracy, harm), yet the policy is a defeasible default any Course Organiser can simply switch off. If AI use were really dishonest and piratical, individual opt-out would be incoherent. The premises overshoot the conclusionâs actual deontic force.
Hidden assumption. The policy presupposes AI use is reliably detectable and attributable. Current detection is unreliable, which raises fairness and false-accusation worries the justifications never address.
Steelman: the strongest defensible versionDrop the external arguments (6â9) and the quality argument (5) as decorative or scope-mismatched. Reframe the conclusion to match the policyâs actual defeasible structure, and add the one strong consideration the document gestures at but never states â administrability and fairness under a collective-action problem.
This is stronger because it (i) rests only on internal, relevant reasons; (ii) replaces âcategorical wrongâ with risk management under detection-uncertainty plus fairness, which actually matches a defeasible default; and (iii) makes explicit the genuinely good bright-line argument â that line-drawing and enforcement costs can justify a blanket default even when some nuanced uses are individually fine (as with speed limits and other bright-line rules).
Even this version remains vulnerable on three points: it must still defend banning disclosed, minor assistance (it leans entirely on administrability and the constitutive-activity worry there); it depends on the contested empirical claim that even scaffolded use displaces skill development; and the unreliability of AI detection undercuts the fairness that P4âP5 are meant to secure.
Want me to develop any one of these into a full reply-and-rejoinder, or turn the steelman into a counter-argument forpermitting disclosed use?
Did you use AI to generate this comment?
They said: âI used Claude with a set of argument analysis skills developed by Marco Meyer to evaluate the policy. I paste here verbatim the result:â
Did you use AI to read the comment?
MB,
gemini says your claude totally misses the point:
BEGIN AI: The second document entirely misses the point of the first by treating a philosophy policy as a standard debate over a technical tool rather than a defense of an existential activity. The first document has two core aspects: safeguarding the internal cognitive process where writing is thinking, and resisting external corporate enclosure of human thought.
The critique misses the first aspect by assuming AI is just a tool on a continuum of assistance. It ignores that philosophy is the activity itself, not a product. It misses the second aspect by dismissing political and ethical objections as mere scope errors, failing to see the first documentâs explicit call for systemic academic resistance.END AI
whatâs the point of the reply?
I hate to say this but I feel like even though we are talking about how much learning matters for students, many philosophers already seem unable to or donât have the patience to evaluate arguments properly or even read.
MBâs intention â based on charitable reading â is to give a playful case study of Claudeâs ability to give feedback (in this case, pretty good) and therefore present more concrete context where people can evaluate the relevance of AI for education in philosophy. MBâs intention is clearly not to appeal to âauthorityâ and to say Boo to the policy just because Claude said Bad. MB invites readers to judge for themselves and perhaps to make more informed decisions.
Edit: typo
Yes, Aliceâs charitable reading is spot on. I am not saying âbooâ to the policy because of Claudeâs response, but rather saying: look here, Claude is producing reasonable results and the policy is saying âban the use of technology in university philosophy educationâ. Now, what are the implications of that? I could go on to post what Claude says here, but most of what it says is covered by this thread.
I, like everyone, struggle on a daily basis to find a workable policy and practice for AI usage. For one, because of the way higher education is regulated in Germany, instructors have a huge amount of discretion in assessment modes. So I am now returning to in-class written exams (closed and open-book), weekly multiple choice quizzes (generated using NotebookLM and Claude), and moderated use: I openly discuss AI usage in my classes and supervision hours with my students. For instance, I have a team of five TAs (advanced BA students, MA students, and pre-doc) for an introductory lecture course in ethics. Every lecture begins with a 10 min multiple choice quiz based on the previous lecture and core reading. I then discuss the draft quiz and the answers and distractors with my TAs and we choose and edit the questions together. They all find that this form of in-the-loop AI assisted learning has raised the standard in several ways: for the students, for themselves, and for our interactions. They are now literally part of the learning-assessment process and learn about how to create quizzes. Why? Because without this use I could not conceivably create such quizzes on a weekly basis and this form of engagement would not be present. And, I used Claude to create the automated grading system (a set of python scripts) that makes the assessment possible in the first place: the paper answers are collected, scanned and automatically land in my email box, then dropped into a folder, and I then just open Claude (Code or Co-work) and write: grade the new file. The students then automatically receive an email with their score and weekly tally. What do the students tell me? They now spend more time studying that in the past!
So it cuts both ways. We need to find adaptive policies and prohibitions are an extreme form and probably will not work in the long-run. One reason for this is that prohibitions generally apply to cases of rights violations and where there are extreme harms to usage to the individual and third-parties â similar to firearms and class A drugs. AI is not like that.
It is perfectly possible to do philosophy with AI assistance. It just depends âhow we do itâ.
Note that if you just pasted my Claude output into Gemini, I am not surprised that you get this result, which actually misses the whole point of my posting â see the charity reading below. But it also misses that Claude steelmanned the position in the first place to take into account this objection. Gemini is just re-configuring the original Edinburgh policy â just as humans do. In contrast, my use of Claude was based on pre-configured argument analysis skills. And the task was only to explicate arguments not come down in favour of one side or the other.
Matthew, I instructed gemini explicitly to say that claude missed the point and to give a reason why for this, one that was sensitive to the two parts of the proposal in the OP. And I told it to do this in less than 100 words because I canât stand long AI posts. Also I primed it by first asking it to consider the greatness of philosophy in the history of humankind and to find the places in its multidimensional space where the idea of fighting for what is of value is represented. So, if gemini saying that claude missed the point was in fact missing claudeâs point, it was me, not my electronic genie doll that is at fault. I realize youâre trying to show what Claude can do with arguments.
In which case Geminiâs reply is quite reasonable because of the prompting (priming). So basically you were using Gemini as an interlocutor for drafting plausible responses. And interestingly, Gemini is somehow reflecting human thinking about this process â which is unsurprising as it is trained on the history of human thought.
Do not get me wrong: AI is a challenge and I agree that wrongly used and implemented it does indeed interfere with our abilities and dispositions of philosophical reflection. But I am not sure it destroys the philosophical mindset nor the products of it.
On the question of writing as constitutive of philosophy which is a premise of the Edinburgh rules, there is an irony to note. In The Phaedrus, Plato has Socrates say that writing philosophy is corrosive on wisdom because people who read philosophy gives people the illusion of knowledge without requiring them to actively learn, internalize, or understand the concepts. Now, if that were true âŠ
One comment is much better than the other one for sure.
Nicolas â
Now this is cryptic. Which one? And for what reason?
Yours is obviously much better. Geminiâs is just a lazy drive-by full of clichĂ©s.
Although the original Claue argument analysis is not a lazy drive by at all. It really depends on how we deploy an AI. Marco Meyerâs argument skills are summaries of argumentation theory.
No, I mean your Claude comment was obviously much better than Runaâs Gemini comment. I was commending your use of Claude.
Thank you, Nicolas. With the appropriate Claude skills and memory logs you can use Claude effectively for targeted tasks. It is particularly useful for complex argumentation mapping, analysis, and re- and de-construction.
Which option do you think philosophy faculty should present to employers, funders, and policy-makers: (A) we forbid students from using AI; or (B) we teach students to use AI to be more efficient, productive, and creative.
A parallel dichotomy: which option do you think makes your students more employable: (A) I never learned to use AI; (B) I was trained to use AI to augment my cognitive skills and to solve harder problems more effectively.
If youâre advocating for (A) in either case, you should prepare to find other work, because youâre probably soon going to lose your job. (And itâs really tragic to see this policy coming from a UK university, since the UK university system is on the brink of total financial collapse. As for philosophy, see Hertfordshire, Nottingham, Kent, Kingston. And itâs paradoxical to complain about âglobal capitalismâ in a policy whose consequence will be that only ultra-wealthy students at ultra-wealthy universities will be able to afford to do philosophy.)
There are ten thousand ways for philosophers (and our students) to use AI to do better philosophy. And for us to teach our students how to use AI in ways that enhance their abilities rather than degrade them. Either we start to develop these ways, or our profession pretty much goes extinct, and that might happen very soon.
How about:
(C) I teach philosophy, not how to use any particular tech even if essential for the modern workplace, whether itâs AI or MS Office or Adobe Acrobat or whatever. And thatâs ok even if you think LLMs will be important in future jobs, because itâs so easy to learn how, and LLMs are getting easier and more intuitive to use as they evolve; even a child can get it to do useful work.
Even if someoneâs too lazy to experiment with AI or just learn from YouTube, there are community classes that will hold your hand in teaching you how to use AI. Not every university class needs to teach AI skills (or even any given skill), and employers, funders, etc. should be glad to know that weâre still laser-focused on our missionâon the things weâre supposed to be teaching (philosophy, in my case).
Anyway, itâs a stretch of the imagination to think that banning AI in any given class or even in an entire department will mean students will be forced to say âI never learned to use AI.â Again, other classes and learning opportunities exist. And why would these studentsâall digital natives, unlike most of usâwant to learn how to use AI from someone whoâs not even trained or an actual expert on how to use AI in the first place?
And itâs even more of a stretch to say âEither we start to develop these ways, or our profession pretty much goes extinct, and that might happen very soon.â This sounds like something an AI would tell you, but you donât have to believe the hype!
Iâm genuinely baffled people think LLMs are easy to use.
Theyâre very easy to use mediocrely!
But itâs very hard to figure out that one isnât using them to anywhere near their possibility, if one isnât getting feedback and instruction and alternate models of how to use them.
LLMs are easy to use? If thatâs what you think, then thatâs exactly the problem. Using them foolishly in philosophy probably is easy. Using them well in philosophy is probably going to be very hard.
Or do you just think that using an LLM means typing a prompt and getting an essay back? If thatâs what you think, then I encourage you to start thinking about other ways to use them (and there are dozens).
Of course, our students havenât thought much about how to use them well either. But the very fact that a philosophy department needs an anti-LLM policy proves (as we all know anyway) that students use them in bad ways. Which justifies the claim that our students now believe that philosophy has been successfully automated. It hasnât, but thatâs what they think. And so do parents, and prospective students, and increasingly taxpayers, funders, and policy-makers.
Thereâs no future for teaching philosophy if people believe itâs been automated. Why would students pay for a philosophy degree? Look at whatâs happened to CS degrees. Codingâs been mostly automated. Itâs becoming a worthless skill. Writing a philosophy essay (which is, according to threads like this, all that philosophy teaches) already has little value and soon will have next to none. The stats for our field are already dismal.
AI will end philosophy unless we incorporate it in positive ways into our craft.
I have to confess, you have pretty much perfectly diagnosed my assumption about what it means to âuse an LLMâ (that is, to type a prompt and let the program do your thinking for you). It sounds like we agree that this is a horrible practice, and that it is also typically what students do, but Iâm very interested in what you say about other, healthier uses of the technology. I admit, Iâm completely ignorant of what these might be, so perhaps it would help if you could be more specific about what you have in mind.
A worry I have is that the bad use of LLMs appeals to students precisely because it allows them to be lazy and offload their thinking. That is the basic problem I think most of us have with LLMs. Personally, I would be thrilled to find a way of using the technology that doesnât involve this kind of laziness, but I fear that making students use LLMs that way will fail to appeal to them for the same reason that traditional methods do. I mean that, because itâs hard to think for yourself and easy to outsource, we would still have to prevent students from defaulting to the âbadâ use of LLMs and force them to use them the âgoodâ way. Basically what I want is for students to learn how to think critically for themselves, but the availability of AI means we canât force them to do so using traditional methods anymore, and I fear that the same temptation to use AI lazily will exist even if we find a better way of using the same programs, as long as products like Chat and Claude are accessible to the general public.
Maybe we have to rethink. If students are being âlazyâ then this seems to suggest that maybe the expansion of higher education as we have it now needs a rethink. At my publicly funded German university, we are facing up to 20% cuts over the next 7 years. I have found myself actually on the side of the cost-cutters. Maybe itâs time to deflate many areas of higher education and reinvest in higher vocational training and in technical and craft skills. It may well be that those who are being âlazyâ are doing so because they have been incentivized to enter higher ed, although they would be better off doing something else but unfortunately those opportunities do not exist. If we are honest with ourselves, a large number of students maybe just buying a âjob market signalâ either in cash and foregone opportunities (fee-paying systems) or as mere forgone opportunities (non-fee paying systems).
We agree entirely. One-shotting an LLM (typing a prompt and getting an essay) is horrible. Itâs mind-killing, and every bad thing you or others have said about one-shotting is true. But one-shotting is not the only way to use LLMs.
Another way to use them is as debating partners in Socratic dialogs. Philosophy has been done through dialogs in the past, and it can be done that way in the future. If you (or a student) can do critical thinking about a topic by reading books or talking with human philosophers, then you (or a student) can do critical thinking about a topic while conversing with the LLM. But itâs not easy to do this â itâs a skill that needs to be taught. A student needs to be taught how to criticize the output of the LLM at each step in a long conversation, how to steer the conversation in valuable directions, that is, how to drive the LLM. Instructors can illustrate this in class.
There are ways to make this work for assignments. The student turns in the text of a conversation, highlighting their parts of the conversation. You donât necessarily need to read the whole conversation, just look at the studentâs moves and evaluate whether the student is pursuing a line of thought. Is the student using the LLM to do deep questioning? Did the student raise objectionis at each step? Likewise, you can just look at the endpoint: did the student drive the LLM to an interesting destination in the space of ideas?
Or you can ask a student to write an essay based on a long dialectical conversation with an LLM, and you can grade that essay really really hard. Wikipedia-level output gets an F. If the student wants an A, they better turn in research-level output. LLMs enable them to do that even in an intro level class.
Iâve tried these procedures in class and on my own, and Iâm persuaded that there are really powerful ways for philosophers to use LLMs (as there are in math and the sciences), but Iâm only just beginning to think about these ways. Thereâs more work to do here.
You can actually track the quality if such socratic dialogues with Claude. After prompting, click on âthinkingâ and you will find Claudeâs âextended thinkingâ which tend to be quite intricate and contain a lot of information without the usual âAI wordingâ. Its much more natural and Claude is actually directly engaging with the enquiry and not just presenting a polished answer. Here is an example. I was testing whether Claude formalize a particular grounding relation (which it did quite well) and then rewrite that in plain English. Notice that Claude recognizes an ambiguity in its own formulations and between the formal an plain English versions. This under the hood processing is worth its weight in gold for developing and testing ideas.
âMatthew is asking me to express the grounding relation as a necessitation claim insteadâconverting the constitutive direction I provided into a statement about what necessarily follows, and he wants both the formal expression and a plain English version of it. Now I need to spell out all the quantifiers, the modal operator, and decide whether this should be a conditional or biconditional. Since the constitutive reading treats political equality as just being the symmetry condition, it should be a biconditional â necessarily, PE(N) holds if and only if E is invariant under all permutations of N, which might even express an identity rather than mere necessity.â
Hold on, I only said LLMs were easy to use, which should be non-controversial. Ask any LLM, for instance.
But I agree that to use LLMs well or masterfully is a different story, which I know from firsthand experiments. And it still wasnât too hard to figure out; it just took time and practice.
Itâs also possible and even likely that LLMs could be used lazily and unproductively, e.g., if approached as a one-shot interaction. (For many of us, this risk is great enough to ban all AI use by students, to close any loopholes and temptation for cheating as much as possible.)
Whatâs missing is the middle path that most people seem to take: all they need is to use LLMs well enough. Not masterfully, but perhaps some back-and-forth exchanges to steer the LLM in the desired directionâenough to get the job done. Not everyone is working on a magnum opus and needs perfection, esp. students.
Itâs not hard to have a conversation with an LLM to fine-tune results this way, is it? If you say it is, then youâre denying the experience of millions of users who have working conversations with LLMs every day.
Anyway, if AI mastery is what you want your students to learn, why canât they figure it out themselves, as you and many others had done? (I assume no one taught you how to do it, right?)
Again, theyâre digital natives and so ought to have an easier time with it. I still donât see why theyâd want or need to learn from someone born in the 1900s who also doesnât have any formal training with AI. (Youâre not telling all philosophers to go out and get this training, are you?)
In experimenting with how to use LLMs well, students shouldnât start with philosophy, since they have little or no context to know what questions to ask or what an error looks like. But they can start with a subject that each student knows very well (e.g., about a hobby) and then keep pressing the LLM until it breaks. This might even be how you learned how to use LLMs expertly, or at least thatâs how I started exploring its limits and capabilities.
If it was easy for us to figure it out, itâll likely be easy for them, too. Maybe they wonât learn how to use AI well for philosophy right away, before they develop some background and skills in philosophy. But the vast majority of your students wonât go on to philosophy jobs anyway, and employability seems to be your primary concern.
Or if it wasnât easy for you to figure it out, can you help me understand why? That wasnât the case for me and, I would bet, many others here. Am I (and many others) the outlier hereâor is it an exaggeration that this is all so difficult that we teachers must teach these skills (despite being untrained in AI)?
So, first off, I donât know that philosophers should be instructing students in how to use LLMs. I think youâre right that we should be focused on the philosophy, and honestly, some people get more out of LLMs than others. I think philosophers who like using them or find them useful should experiment with ways of incorporating them into teaching, but it would probably be counterproductive for people who donât like using them to try.
The second thing is, the stuff that it is easy to learn to use this stuff for is not very useful, or at least has pretty limited usefulness. One way to think of it is like this: in a way, Excel is extremely easy to use. If youâre like me, the primary thing you use Excel for is keeping track of student grades. That takes a little bit of learning. But you can learn that incredibly quickly. So again, you might say Excel is really easy to use. In a sense it is, if you only use it for very basic tasks.
The quality of the outputs you get with an LLM can vary considerably depending on the quality of your instructions; whether you provide them with templates or examples; how you break the task down for them; learning what they will misinterpret and what they wonât; etc. This is one reason why people have very different experiences with their capacities. I think people who think they are easy to use are getting them to do fairly basic stuff, and then try to get them to do something more sophisticated, get a bad output, and then conclude it is just something the LLM canât do. But what they can do (at least now) is very much driven by the instructions. So when you say theyâre easy, either you are much smarter than I am, or you just arenât using them for very much.
The people who are really using this well talk about building out (in code) customized interfaces for instructing the LLMs. Thatâs well beyond anything Iâm doing. That said, since Iâve learned how to create and use customized GPTs, Iâve found that what these things can do is much more useful and much more impressive than what I could get them to do a few months ago. (Though 5.5 is also a big improvement of 5.2, and that explains part of it.)
Iâll also just add that I donât find the digital natives stuff persuasive at all. With a few exceptions, the kids these days suck at using technology.
In the empirical sciences students are instructed to use mathematical and statistical software packages and I recently spoke to a well-published mathematician who said the entire proofs contained in the appenedix of his paper were automatically done by Codex. And it is now standard practice for those working with simulations in the empirical sciences to have the coding done by an LLM. So why should we hold back this technology from students of philosophy where it is appropriate? For instance in formal logic, but also in complex argumentation analysis â to build argument trees.
I am starting to believe we just need to systematise the training in LLM usage and come to some shared understanding of best practices of assessment.
You could be right. I donât want to express any strong views on what ultimately will be the right and wrong ways of using the technology in philosophy. I just also think that there are aspects of using LLMs that are easier for people with some background in coding. I also think that philosophers could be uniquely positioned to use them well, since we are trained to use natural language in unnaturally precise and structured ways. But my general thought would be, at this point, we donât really know what the technology is capable of. Its capabilities are also changing very quickly. (I can certainly get outputs much closer to what I want with Claude, even with pretty mediocre instructions, if I am just willing to max out the capacities and burn my weekly usage on the project. So maybe these points Iâm making about it being hard to use will seem silly in a few years. I donât know.)
The basic issue right now, I think, is that most of us donât understand how to use it very well. A lot of people in the field clearly find it extremely unpleasant to interact with. (Which is understandable. Chatâs default personality and prose style are really annoying.) And I just think that asking people who donât get it and hate it to teach with it is going to make the teaching worse for the students.
I think it probably is a good idea, though, for each department to have someone who likes using it and is encouraged to tinker with it, to try to figure out what ways of incorporating it into teaching work and donât. I just donât want to say, at this point, that this should be mandatory for everyone. And I also think we should we be open to the possibility that traditional ways of teaching philosophy help teach skills that it is easy to fail to develop if too many of the assessments are based on LLMs. So it might be best if students are exposed to a variety of teaching and assessment methods, and instructors should adopt those they feel most comfortable with.
But personally I think what youâre suggestingâusing Claude to build argument treesâis very cool. If youâve posted anything online about this, or donât mind if I email you, Iâd like to learn more about how it works.
Sure. I have just pmâed you.
It took me awhile to figure out that I could engage in dialectic with the LLMs. Specifically, that I could criticize, it and challenge it, and demand that it give reasons. I didnât see it as a partner in argumentation. (Initially, I too was entranced by the one-shot use case.). And of course the early LLMs werenât very good at reasoning or holding conversations.
After much trial and error, I realized that you can teach critical thinking by having students criticize the LLMs. (Iâm not talking about criticizing hallucinations â though thatâs necessary too). LLMs often give generic or shallow answers (the least common denominator response). Those are open to criticism, but that needs to be taught.
And I think arguing against an LLM is a fascinating process. The LLM gives some generic answer, then your job is to persuade it that itâs answer is wrong or just too shallow. I suspect (tho I donât know) that it might be easier for students to argue against an LLM because itâs not a human and they donât worry about offending it.
I think sparring with an LLM can teach lots of logical skills.
Thanks, Eric (and Derek and Kenny in other comments about this).
What I think Iâm hearing is that LLMs can be very tricky to figure out how to use well. I agree.
Still, all of us here were able to figure it out without âgetting feedback and instruction and alternate models of how to use themâ (Kennyâs comment), right?
Assuming we philosophers arenât special in this regard, and many non-philosophers have also figured it out, why canât students also do it, esp. if trial-and-error is one of the best ways to learn something? Students would seem to be the most motivated to learn this, if their future survival really depends on this skill.
But sure, teaching can be more efficient than a self-study. This is especially so if thereâs a long history of ideas or experiments (trial-and-errors) to work through, although thatâs not the case with LLMs.
Still, how difficult is it to distill our hard-earned skills into a short lesson, even a YouTube video?
E.g., Eric said, âI think arguing against an LLM is a fascinating process⊠I think sparring with an LLM can teach lots of logical skills.â
I agree with that, too. But I can demonstrate the basic concept in a 5-minute lesson, and students can go off and practice that on their own.
Eric also said, âIt took me awhile to figure out that I could engage in dialectic with the LLMs. Specifically, that I could criticize, it and challenge it, and demand that it give reasons. I didnât see it as a partner in argumentation.â
Yes, I agree thatâs a critical breakthrough and an essential insight that might not have been obvious in the early days of LLMs. But itâs well known now, isnât it? Or if itâs not, you just explained the basics of that lesson in 3 sentences!
Derek hit the nail on the head with this comment: âmaybe these points Iâm making about it being hard to use will seem silly in a few years. I donât know.â
I donât know what the future will bring, either, but AI fans keep telling us that this is the worst AI will be, and that AI will only get better and better over time. That would presumably include being more intuitive to use well and more accurate in their results.
If so, it seems very possible and even likely that LLMs will be able to do great work with a one-shot prompt (or with much less hand-holding) in the near future. In fact, it seems the only way to believe LLMs will still be hard to use in the future is if you donât believe that they will get better and better over timeâŠ
Or if these lessons really are so hard to impart, how much time exactly are you taking away from teaching philosophy, i.e., what you were hired for?
If itâs a significant amount of time, perhaps itâd be better to have a dedicated 1-unit standardized course (or at least pre-vetted courses) for these lessons, than to take away that time from what weâre supposed to be teaching. If we let rando and untrained faculty teach AI literacy however they see fit, students could get that AI instruction over and over and over each academic term and in possibly conflicting ways.
On a related note: if LLMs are so flawed that it takes a ton of manipulation to make them work well, can they really be the huge threat to jobs everywhere that folks are making them out to be?
How trustworthy could those results be, if we need to so carefully manipulate the LLM? And who are all those people who somehow have this rarefied, job-destroying knowledge, if they hadnât been taught it?
Or are you thinking that these people are currently students of other instructors who are willing to teach them AI skills, and you donât want them to eat your studentsâ lunch?
Ok, thatâs fair, but then youâre preparing your students for a miserable future where thereâs fierce competition for a dwindling number of good jobs, thanks to AI displacement. Itâs possible the future may turn out that way (complete with UBI), but itâs also very possible it wonât and that AI is still overhyped.
For my part, I wouldnât want to present that miserable future to my students as a certainty or even a likelihood. Maybe some students would be motivated to work harder so to not starve or be homeless in the future, but Iâd expect this would just cause more anxiety and de-motivate most students.
Anyway, I and many others donât think the future requires AI. That opinion is fine, and youâre entitled to your own opinion, too, esp. since no one knows how all this will play outâŠ
Good luck to all of us, and have a nice weekend. â
Thanks for this reply.
About this point:
I agree. I seriously doubt that theyâre a threat to jobs everywhere. I think that certain professions may see serious job loss; but overall, if this technology is useful, that means it will make human labor more productive, which should increase demand for that labor. This isnât guaranteed to happen, but it is at least as likely as the more nightmarish outcomes people talk about all the time. And if you are thinking about it this way, the issue isnât that students in other fields will our studentsâ lunch (because itâs not a zero-sum game), but that people who can use LLMs to enhance their cognitive skills rather than a substitute for such skills will be better problem solvers, better at teaching themselves, and better at critical thinking (given the new information environment we inhabit) than they would be otherwise.
My experience is that the answers are as generic as the question itself. If you are specific, and tell it to be specific, the output generally is specific. If you target an argument in a text in a specific way, such as âreconstruct and test if this argument is modus ponens or modus tollensâ by presenting in premise conclusion form; or âdetermine which argumentative fallacies are present here in this textâ, an LLM will generally perform quite well (especially if it has, like I have with Claude Argument Skill pre-loaded). You can then look at the structure and go back and forth with the LLM to justify the reconstruction, asking it for textual evidence for the premises. You can then even go one further and request reconstruction in symbolic logic etc. Basically, anything a human can do under instructions, an LLM can do.
As people have pointed out on this site over and over and over already, if a person does not already have the relevant cognitive skills and excellences, they are not going to enhance any of the relevant abilities by using AI shortcuts. (Certain abilities may indeed be enhanced, but not any philosophy-related abilities. Philosophy-related abilities include many of those that make human beings at all interesting in comparison to other animals.)
Supported academic philosophy will go extinct either way as long as we continue on the current (econo-political) framework for AI development and the current culture wars involving higher education, while at the same time not developing counterbalancing national and international regulations. So itâs really no argument to say that philosophers should go along to get along.
Iâm not talking about âusing AI shortcutsâ. Iâm talking about using AI, as I said explicitly, to enhance and augment our skills (and those of our students). The problem is that you canât see how thatâs possible. Scientists can (and do), mathematicians can (and do). But philosophers apparently cannot. And thatâs tragic.
Eric, when you think of examples from science and math in which AI is used well, I wonder what kinds of AI are involved? Machine learning AI has been being used in science and math for quite a few years before generative transformer genie dolls were released upon the public. I suspect many of the most impressive examples of use of AI in science involve machine learning, but not generative transformer AI, though obviously there are impressive discoveries here too, as in: the detection of a single cancer cell in otherwise normal patterns. In any case it is the generative transformer genie dolls that are problematic from my perspective, and specifically problematic when used to mediate or replace human interactions and relationships that make culture possible, such as those between a teacher and a student.
A: I teach students to think methodically, charitably, and creatively.
But, you know: I dare you to replace me with a chatbot. I am entirely confident that Iâm better at everything I do than it is, and I bet that, ultimately, Iâm more efficient at it, too. I also bet I outperform my human + chatbot replacement.
Did I say anything about replacing humans with chatbots? No, I didnât.
You said I should prepare to find other work because I was going to lose my job.
There are certainly better teachers and better researchers than me out there. But of I lose my job, my institution is not likely to be hiring one of them to replace me. And, as I said, Iâm confident that Iâm better along just about every metric (save sheer scale when it comes to feedback; I will concede that 200 students a semester is just about my limit) than my chatbot or chatbot-enhanced replacement.
If youâre going to lose your job (teaching philosophy), that doesnât imply youâll be replaced by a chatbot. More likely, you wonât be replaced at all.
I donât think youâre more efficient at it. The AI is efficient enough at having dialogue with students that it can have one-on-one conversations with every single student every night at a cost to the companies of likely under $5 a month (and $0 for the students).
But Iâm sure you are much, much more effective when you do have dialogue with them, and youâre definitely not replaceable for this.
Well, Iâm sure you could replace me, and standards would rise! But nobody here will make you an offer youâd struggle to refuse.
I hope I have not transgressed any ethical boundaries here, but I used Claude in Chrome to summarise the status of the discussion (121 comments), just to get an idea of where the central tendencies lie. I post them here, again verbatim. Note that this is what an LLM is ideal for: summarizing complex textual content so we can start to make sense of the overall picture. If I have transgressed by doing this, I ask for your apologies and that the moderators either delete this posting or not post it at all.
Here are Edinburghâs nine justifications, one line each (paraphrased):
The common split in the comments: 1â5 are the internal/pedagogical reasons (widely seen as the strong ones), 6â9 the external/ethical-political reasons (widely seen as overreach for a course-assessment rule).
What does it for you, makes you weaker: https://youtube.com/shorts/oIebgNXDy2U?si=TSkyPnkNDGqPijYp
Not sure if itâll work (or if it does how often it will), but hereâs a gift link to a relevant NY Times piece on the promotion of AI literacy in the Cal State system: https://www.nytimes.com/2026/06/01/magazine/ai-university-college-california.html?unlocked_article_code=1.m1A.CIRQ.duybCKODkpiJ&smid=url-share . Philosophy professor John Sullins of Sonoma State is quoted near the end.
I think this sorry tale and the Edinburgh policy are both examples of ideological extremes: âtech utopianismâ and âhumanist utopianismâ. Common sense is enough to tell us that neither are practicable. Blanket bans and blanket permissibility. From the discussion here, I hope that what we learn is that we need adaptive strategies which relate to the particular educational goals we have in mind for a philosophy degree.
Sure, some instructors may be looking for middle ground here. And they may be doing so for any number of reasons, incl. that they want to appease the powers-that-be, or maybe they really believe theyâve discovered the path for responsible AI use.
But where are these adaptive strategies youâre talking about? Weâve had since Nov 2022 to develop them, but I have yet to see anyone crack that nut. Just telling people we need them isnât going to make them suddenly materialize.
Even if some instructors are ok spending (uncompensated) time in overhauling their lesson plans (which may be bespoke to their courses), they too have an enforcement problem, just like those whoâd want a full blanket ban on AI use:
If we canât enforce an AI ban, then how can we enforce a limited AI ban on certain uses to keep students on the path of responsible use? Short of requiring students to write things in person and by hand (which is itself a limited AI ban, and even that can be gamed), it seems any âadaptive strategyâ can also be abused. And this was a chief complaint about AI bans.
If anyone has an abuse-proof assignment or lesson plan we can inspect, I would love to see that.
But it seems that in none of these optionsâwhether blanket ban, blanket permission, or something in the middleâcan we save everyone. Students who donât want to work or who are determined to cheat can likely find a way to do that regardless of your pedagogy.
Considering that this middle path would require the most work, and that no one knows how the future will play out (whether AI will be as important as some claim), itâs not obvious that thereâs one right path to take here, even if I have very strong opinions about it.
Yet some folks somehow have so much conviction that they know an AI ban is the wrong way. But I can even imagine that blanket permission to use AI could be rational, e.g., instructors who have been so beaten down by AI pushers and too tired to resist or overhaul their courses, esp. if not given real resources or help from their schools.
Anyway, I would love to see some good âadaptive strategiesâ, and all universities should be giving their instructors those sample-assignments, which should be as pedagogically valuable as whatâs lost. But they donât, and Iâm not even aware of a single university that has.
Again, if anyone here thinks they have it worked out, or can point to a bank of good sample-assignments, please share with us. People are still asking for this help, but clearly itâs very much easier said than doneâŠ
Note: all this puts aside broader social and ethical issues with using LLMs, esp. in a learning environment, which really shouldnât be ignored.
Well, what the strategies look like are going to depend on the academic system we are working in. Currently, in the German system a blanket prohibition is legally impossible and insturctors have a huge level of discretion. I have given up on seminar presentations, instituted in-class tests. Luckily term papers and theses are closely supervised so I have an AI discussion with all my students and we discuss AI usage: what is permissible and what is not. From next semester, I will add a written exam for seminar courses that will form part of the assessment combined with the term paper. That gives me a baseline to assess the student and how much is AI influenced. These are intensive and not scalable to large number of students.
So, I doubt there is a single âadaptive strategyâ but rather a range of best-practices that are going to be institutionally localized.
Another strategy is we have to move away from generic questions. For term papers I have started to focus on very detailed analyses and do not accept or assign literature reviews anymore.
Ah, I apologize if I had misunderstood you earlier, Matthew.
It now sounds like you are talking about how to cope with AI (i.e., how to AI-proof as much as possible, esp. when a ban isnât allowed), as opposed to adapting by integrating AI into assignments.
I had thought you were talking about the latter, and I was challenging that as a possible fantasy, at least to do it well enough so that the adapted course is as pedagogically strong as before. Thereâs no substitute for struggling with all the work yourself, whether itâs in the gym or in the classroom. Also, enforcing guidelines for permitted AI use may also be very tough, just as with a full ban.
But if you were really talking about the former, I agree that instructors need more coping strategies. Many of us are just limping along, shell- shocked, as the walking woundedâcasualties in someone elseâs war.
Still, a blanket ban can be a good option, if allowed. Even if enforcement is a problem, the policy can still be a great deterrent, not just because of the penalties but also by forcing the student to make a deliberate choice to cheat.
Just as with DNA testing, itâs also possible that reliable AI detection methods or audits will be developed in the future that can be applied retroactively. So, students may be deterred by the prospect of having their university degree rescinded in the future, just as some academics have lost their PhD degrees and jobs when plagiarism was later found in their dissertation or other work.
Anyway, this is generally what society does for the problems it faces. E.g., we want to stop cyber crimes, but enforcement of cyber laws is still a huge challenge, esp. attribution of the attack. That doesnât mean we should surrender to the problem, but we cope as we can, until we develop better methods (e.g., precise attribution for cyber crimes, or DNA testing for murder cases).
De-criminalizing those problems may solve the enforcement problem and appease the loud voices pushing for it, but itâs more likely to create other, worse problems. Similar case with âde-criminalizingâ academic cheating, Iâd argue.
Using AI in the workplace is a different conversation, where product is more important than process or learning and where dishonesty norms are different. So, even if some instructors find value in using AI for their work, that doesnât mean itâs valuable for students to use it in their learningâŠ
âShort of requiring students to write things in person and by handâ â thatâs in fact it. Other than in small seminars with senior majors, everything that is done for a grade is done in my presence on paper. Itâs virtually cheat-proof, and cheating attempts are super-obvious and easily discovered.
âŠsome inaccuracies @ start of article. but on my csu campus, yes, phil is now fully anti-tech anti-ai populist. many of my colleagues swayed by bernie throwing in with yudlowski (âweâre all gonna dieâ). the faculty are now reinforcing what the church crowd is cultivating among the students.
And to think this is only the first swing of the wrecking ball. Just wait for what comes next.
Some of the comments on the story are interesting â one gets to see what the new talking points will be by those who want blindly to submit to the integration of this extremely powerful technology in all our systems for before there is time to measure its effects on human psychology.
The same was said of the steam engine; the same was said the internal combustion engine; the same with personal computers etc. I use Claude every day for teaching, research, and administrative purposes. I design, experiment, and test assessment procedures that a year ago were unthinkable and are raising the floor and not lowering it. I engage with literature in a way once unthinkable: for instance, I used Claude to assemble a bank of examples used by Scanlon and others in discussions of his âreasonable rejection testâ and then fed them back into Scanlonâs original analysis only to find inconsistencies that I previously overlooked. I can explore ideas about, say, political equality, across a range of formal methods and structures that would have taken weeks in the past, and now is a matter of a day or two. I can create individual teaching content and in-class exercises and presentations including simulations, once unthinkable. All this raises the bar. And with it we must take our students. For those working in highly analytic and formal philosophy, AI is super-power in the same way it is for those in the empirical sciences.
One further thought: maybe one of the effects of this âextremely powerful technologyâ is to stimulate a new kind of âhumanismâ. Maybe we will appreciate human crafts even more: maybe it will stimulate more inventiveness on how we distinguish ourselves as âhumanâ. I am not suggesting blind deference, but I remain unconvinced by the doomsters. Just think, every generation seems to have its apocalyptic scenarios, none of which ever come to pass. Maybe as philosophers, and those in the humanities in general, have an over-inflated sense of importance of the contributions we actually make.
How it works
Once you click Generate, Ollama reads this article and crafts 5 comprehension questions. Your answers are graded against the article content â general knowledge won't be enough. Score 70+ to count toward your certificate.
Questions are cached â you'll always get the same 5 for this article.