Should Humanities Faculty Aim to Ban AI from Their Classrooms?
“Because of the ubiquity of AI technology, students will likely be using it persistently outside the classroom in their personal lives. The humanities classroom must be a place where these tools for offloading the task of genuine expression are forbidden—stronger, where their use is shunned, seen as a faux pas of the deeply different norms of a deeply different space.”
[“Die” by Tony Smith]
Bans on the use of AI by students were described as “madness” by D. Graham Burnett (Princeton) in a widely circulated article in The New Yorker (see this post). But Fritts thinks that universities should seriously consider this “radical response.” She says:
Preserving art, literature and philosophy will require no less than the creation of an environment totally and uncompromisingly committed to abolishing the linguistic alienation created by AI, and reintroducing students to the indispensability of their own voice.
Fritts has been serving on a pair of committees her university has convened for the purposes of responding to artificial intelligence, especially its use by students. In thinking about her role on such committees, and the role of humanists in general on them, she suggests that it can a place to join together and push for that “deeply different space”:
This is precisely where humanities faculty on AI committees can make a difference: these radical policies will never be given the time of day by university administrators unless we in these disciplines can present a united front regarding the true aim and importance of a humanities education. It is risky, I acknowledge, to admit that our departments are not in the business of producing new products or supplying students with expertise that will increase their earning potential after college. But at a time when higher-education funding is on the chopping block, the prospect of AI will make the alternatives even riskier. If our deans and boards of directors think that our primary goal is to produce—arguments, manuscripts, essays—then the ways in which we are deskilled and de-personed by AI will have no obvious negative impact on meeting these objectives.
Humanities faculty on AI committees must resolve to be honest about what is at stake in these policies. We must not shirk our own duties of authentic self-expression, settling for a watered-down compromise that seems humbler, safer or more “serious.” If AI is allowed to expand its presence in the humanities classroom, I will put my money on the bleakest predictions of education prognosticators. Given what is on the near horizon, it seems to me that, even in the riskiest setting, we stand to lose relatively little by being bold and honest about the true nature of our work.
You can read the full piece here.
(via Frank Cabrera)
I support this proposal for reasons associated with Fritts’ comments above.
Thinking is materially realized. From Socrates quizzing the slave in Meno gesturing to a square drawn in the sand to students using LLM chatbots to make sense of complicated texts, our thinking is scaffolded by material systems. The character of the material system partially determines how we think, and perhaps even what we think. Some material systems force slow thinking – writing with a pencil on paper, for example – and some facilitate rapid, gappy thought – reading text produced by an LLM chatbot, for example. Additionally, some material systems generate distraction. Writing with pencil on paper moves us towards doodling. Thinking with a screen on an internet-connected device moves us towards flicking through different websites or, in the case of an LLM chatbot, asking random questions that just occur to us.
Distraction comes in different forms. It can be a sort of gentle movement away from what one was attending to, as when one slowly drifts into reverie or it can be abrupt, as when one scrolls social media or surfs the web.
My view is that philosophy is a slow process that benefits from slow distraction. It is also a discursive process that requires attending to the person or people with whom one is communicating. If that is correct, all internet-connected screens are threats to doing philosophy in its ideal fashion. (I once heard Cornel West describe the internet as a “weapon of mass distraction” and philosophy as the best defense.)
But, ChatGPT and the like can be very useful for producing philosophical texts. They are useful tools for marshaling citations and chasing down other texts relevant to one’s work. But, to some degree, this use case only applies once one has developed the “slow thinking” skills essential to philosophical reflection and discussion.
So, I think that there is a good case to ban LLM chatbots – and screens more generally – from a classroom. There is a good case for destabilizing students’ habitual reliance on internet-connected systems as the material scaffolding for thinking. Productively disrupting habits of thought is probably one of the ideal outcomes of philosophy. Forcing students into screen-unmediated boredom (as opposed to screen-mediated boredom) may be good practice at least in part because this disrupts the habits of thought into which our students have fallen.
Quick side-note: we might distinguish AI from LLM chatbots. AI is a fairly broad category, and it includes many capacities unrelated to the philosophy. The concern we have is with the widespread use of chatbots.
Really good points – though they seem to me to suggest a range of classrooms where we require everyone to work with a specific modality, rather than all of them being pencil on paper or chatbot-free! (Maybe most of them should be though – I haven’t thought through what balance of modalities is best.)
Yes I agree. Heterogeneity of methods is good. I just think that for most students, especially non-majors, one of the methods, using or permitting use of LLM chatbots is among the worst methods.
Agreed. Build an ark, while we still can (Noah was a conspiracy theorist — and then it rained).
The humanities classroom is a place where these tools should be used intensely, and students should be taught how to intensely use these tools.
I plan to make a short video named “How to Use AI to Make Yourself Smarter” for my fall classes. I’ll put one of the reading questions into AI and then explain how to investigate whether the answer is gives is correct.
The point, as I will explain to students in the video, is to help orient them if they are feeling lost about how to approach a philosophical text. I don’t expect it to eliminate lazy attempts at using AI – that’s what well-phrased exam questions are for – but I hope it will make it easier to understand the readings in ways that will help diligent students do better on the exams.
All these things are doing is conditioning students for enslavement by robbing them of the opportunity to develop their own critical faculties. Anyone who actually cares about his or her students should be explaining how absolutely anti-human this technology is.
Welcome to the age of mass illiteracy and the most effective propaganda apparatus you’ve ever seen. Stalin would have loved this.
For their final projects this semester, some of my students used AI to generate nice illustrations, others to generate a template for a debate, yet others for podcast scripts, all of them, at least as far as I can tell, with their own thoughts and work in both the input and output. AI allowed them to embark on more ambitious projects and not feel overwhelmed by the task. Of course, some students are also getting away with illicit uses and lazy offloading of their own skills. Still, I didn’t feel like allowing them to use AI in clearly circumscribed ways was contributing to their enslavement. I hope not.
When you say that *they* embarked on these more ambitious projects, what do you mean?
I’m not sure what you’re asking. My students worked on and presented projects that were enhanced by GAI in some ways. They used a tool to achieve, quite efficiently, something they probably would not have achieved otherwise or more poorly or at greater cost.
My question was about the degree of their involvement in actually producing the project materials. So, you mostly answered it (thank you).
That said, my follow up is this: do you think that their use of AI for these things undermines the degree to which they actually contributed to the project–and does that matter? I don’t mean to prejudge anything. I’m genuinely asking what you think.
For example, you mention that they generated illustrations. I’m sure that their projects were more sophisticated because they were able to do this. But, clearly, *they* weren’t making the illustrations in any traditional sense–they were crafting prompts and letting the AI take it from there. That’s not necessarily bad. I’m just wondering whether that counts as “their achievement”–and if not, whether that matters.
(I’m granting that even if your students wouldn’t count as achieving illustrations and the like, they might have used them to achieve things on their own–e.g., arranging the illustrations in an informative way. But that’s partly why I asked what you meant by their embarkment.)
“do you think that their use of AI for these things undermines the degree to which they actually contributed to the project–and does that matter?”
No. Not in a sense that matters.
Is that a “no” to both questions, or just the question of whether it matters?
Both.
it’s not clear you are even entitled to apply the concept ‘achieve’ here without negating the concept as it has formerly been used.
Why? Have you heard of the concept of tool?
do you think a faucet is a tool? that a new kind of faucet can be invented, and that it can pipe strings of words and imagery into the works ostensibly made by students, would still not make uses of the faucet instances of achievement.
I’m going to assume you have not read what I wrote.
A pencil is a tool. When someone uses it to create an illustration, they—rather than the pencil, or no one—achieve the production of the illustration. I assume that you agree with this based on your last comment. I do too.
An AI is a tool. When someone uses it to create an illustration, they—rather than the AI, or no one—achieve the production of the illustration. Do you agree with this?
I’m open to the idea that the person achieves *something* in the latter case. But it reaaaaallly seems (meaningfully) different than the former case. If the latter is just another example of achieving something through tool use, as you seem to be suggesting, what would you say in response to the apparent differences between these two cases?
This is getting bizarre. I did not claim that the achievement was the illustration? I’ve mentioned final projects and emphasized that they were essentially informed by the students’ input, such that even with enhancements (e.g., a picture in a slideshow or a nice intro to a podcast), the output was still theirs. I like running races in light and fast shoes. I need certain kitchen tools for certain recipes. The runner is me, the cook is me. These are tools I use to achieve my own goals.
In my original comment, I wrote:
In case this wasn’t clear, neither the illustration, nor the script, nor the template were the goals being achieved. You guys should try charitable reading, I teach it to my students, and it works!
“You guys should try charitable reading, I teach it to my students, and it works!”
I’m not attacking you, there’s no need for disrespect. At any rate, this is precisely why I made multiple requests for clarification before asking my questions. A recap:
You initially said, among other things, that “some of my students used AI to generate nice illustrations.” You clarified afterward by saying, in part, that “they used a tool to achieve, quite efficiently, something they probably would not have achieved otherwise or more poorly or at greater cost.” This still didn’t really clear things up, since I remained unsure what the “something” referred to (the overall project? the parts of the project?). In any case, I then replied with a comment which explicitly granted what, I think, you’re now making explicit–that perhaps the illustration wasn’t the goal, but a means to a goal:
“(I’m granting that even if your students wouldn’t count as achieving illustrations and the like, they might have used them to achieve things on their own–e.g., arranging the illustrations in an informative way. But that’s partly why I asked what you meant by their embarkment.)”
You didn’t address this in your reply. All of this suggested to me, sincerely and with a charitable intent, that the illustrations counted. If they didn’t, then great–that answers my question. A simple “no, I wasn’t counting illustrations”–or even a more productive “here are some things I was counting and why I think they’re like the products of tools”–would have sufficed.
It’s not getting bizarre, don’t worry; it’s just a miscommunication. I tried to avoid this miscommunication by repeatedly asking what you meant and bringing up the possibility that you weren’t counting illustrations as achievements. Part of being charitable is reciprocity on the part of your interlocutors—answering their clarificatory questions, addressing confusions when given the opportunity, etc. I hope you teach that to your students as well (I often don’t; shame on me).
For what it’s worth: great, if “achievement” just referred to the overall project, then (although I’m still skeptical of this claim) that clears things up for me, thanks.
Sorry, I think this was mostly in response to An Adjunct’s comment. But you’re right that charity goes both ways. My bad.
To put things back in context, I was initially responding to an emphatic comment about enslaving students in ways that Stalin would have been proud of. I noted that some limited, controlled uses of AI (so, something short of the radical policies proposed by Fritts) need not lead to such extremes. That’s about it.
Fair enough, I also apologize for my hostile response. And yes, I certainly agree with your response to that initial charge of enslavement!
Belatedly jumping into this thread, pencil and AI are both tools. The fact that we value the learning outcomes that can be achieved by (and may in fact be heavily dependent upon) using a pencil as a tool does not mean that we must value all other learning outcomes associated with all other tools.
This catastrophizing won’t age well. Moreover, we have a duty to our students to prepare them to thrive in a world of AI. We should distinguish between reliance and competent use. You can already converse productively with AI about issues in humanities. I see nothing wrong with having such conversations in a classroom so long as we know it’s AI and that it’s fallible, as are we all.
I would be happy to wager that the catastrophizing would age like fine wine, except that this seems so obvious to me that anyone who disagrees at this point seems like they must be evaluating the situation with such a different set of values than me that I can’t imagine both of us coming to agreement on what a future would look like in which I am right vs. a future in which the catastrophizing does not age well.
I look around and I already see some students completely unable to think for themselves, let alone turn their thinking into writing via the process that we all know (or that I thought we all knew) issues in great dividends for the thinking itself, since having to write one’s thoughts clearly often leads to the clarification of the thoughts themselves. The proximate cause of this is obviously AI. Of course, many other factors play a role: that students are inclined to instrumentalize the value of their education, that they face many demands on their time, etc. all don’t help. But AI has made the problem many orders of magnitude worse.
I once had a student who wrote a senseless paper because the paper’s central claim was generated by AI. It made grammatical sense and had the shape of an idea but two second’s worth of thought would clearly show it was nonsenical. But the student didn’t write the paper with AI, I think, and I was reluctant to say to them “the computer has fed you nonsense, stop trying to defend it,” because I was worried they might’ve thought it up themselves and that I would be insulting them. (I’m sure my hatred for AI causes me to overestimate how much a student would resent being mistaken for AI, but still this and also the uncertainty hold me back from accusations even today.)
The student came into office hours (digitally) to discuss the paper, and they kept saying such strange things in defense of the point that I could not understand what was going on. I would clearly explain why the point made no sense and they would respond with some weird digression. Looking back on the interaction I realize they were typing what I said into ChatGPT and reading me back the answer. This should have been obvious to me at the moment, but it simply hadn’t occurred to me that anyone would do such a thing, and so I thought the student was just typing down my points or their thoughts and then taking a moment to think before replying to me. This was of course not just a waste of my time and the student’s time, but also of their intellectual capacities, which at no point in the entire exercise were ever engaged.
Now, of course, that AI is sometimes used badly need not entail that it is always used badly. But it seems ludicrous to me that anyone can ever be prepared to thrive in a world of AI, as you put it, without at least minimal development of their faculties in a world without AI. Otherwise it’ll be no use for them to keep in mind that AI is fallible. They will not be able to distinguish the cases in which it fails from the cases in which it doesn’t. And it’s for this reason that I think the catastrophizing will age very well. In part because of people like you, who promote responsible AI use, and in (larger) part for dozens of other reasons, irresponsible AI use is going to smash the world up. Whether we can put the pieces back together afterwards remains to be seen.
This is an excellent point.
People who say that we shouldn’t “catastrophise” seem to assume that a prediction is being made according to which people in the future will agree that things went badly.
I don’t assume that at all. I’m perfectly willing to grant that people in the (near) future will think that AI has been a great boon to the humanities and that those who said otherwise were just overly fuzzy fuddy-duddies and Luddites.
I am worried that the values of these future people will be the wrong values and that they will not have the capacity to see that.
Here’s what I mean: I think that within just a few years most of the people who want to ban AI from humanities classrooms will have had enough experience with AI to realize how they can be used productively in class.
I think one can accept that a tool is used productively in some classes, while other classes might also productively ban that tool!
People who want to ban the tool from all classes are probably catastrophizing in an unhelpful way. People who think the tool should be available in all classes probably also aren’t thinking helpfully.
Kenny, I agree. Instructors should be free to choose whether to use AI in their classes, and there are contexts in which it should be prohibited and others in which we should be free to incorporate it.
But a ban on AI in all humanities classes strikes me as just a wee bit reactionary, closed-minded, and dare I say authoritarian.
If I ban students from using a cheating service where you pay other people to write papers for you, is that authoritarian?
No but that is not what James is talking about?
What sets the two apart with respect to whether they are are authoritarian?
A total ban would apply to a large spectrum of AI uses that are not even remotely like the cheating analogy you offered. Most of the uses of AI that James and others are condoning are not relevantly similar to this strawman of total outsourcing. I (and I assume James) think this should be prohibited. But that’s not the scope of a total ban!
I don’t think it matters whether cheating is or isn’t similar to using AI. I could have used a ban on eating food in the classroom or a ban on talking without raising your hand or whatever as the example. My question is just whether banning some things is authoritarian and banning other things is not, or whether banning anything in a classroom is authoritarian.
Oh. No, not banning anything. But the urge to ban more and more things, and to ban with a wide scope, yes, I’d say can reflect some authoritarian tendencies!
Is banning 20 cheating agencies all of which offer to write your paper, or 20 different kinds of food, more authoritarian than banning one?
Is banning AI in addition to cheating agencies as much of a reflection of authoritarian tendencies as banning eating in class in addition to cheating agencies?
In general I guess I’m just having a lot of trouble understanding my desire to ban AI as authoritarian, at least insofar as calling something authoritarian is supposed to tell us anything interesting about it. I ban students from standing on the tables and shouting about their favorite TV shows, but I don’t think this reveals anything untoward about me. Ditto for banning AI use in class.
I can’t speak for James but yes, if your first reflex is to ban a lot of things that don’t need to be banned, because they can be handled by incentives and other policies, then you might have authoritarian tendencies. For example,
“I ban students from standing on the tables and shouting about their favorite TV shows.”
I don’t ban any of those things nor do I think I need to. I try to create a class culture in which these things just aren’t done. I ban only that which 1. would be a clear impediment to learning if it happened and 2. is actually likely to happen in the conditions I’ve created.
But maybe you ban a lot of things that need to be banned, that’s okay, I do too. And maybe specific bans on AI use are needed, I do that too. My only point was that James clearly seemed to have something different in mind than any ban = authoritarian. I think the urge to ban as a response to problems is generally bad but that’s because of my own libertarian tendencies which you’ll probably see as a vice.
Maybe we are using “ban” differently. I of course have never explicitly said a single word about whether anyone ought to stand on tables and shout about TV shows. By “ban” I just meant that I would not tolerate it if it were to occur. If that is my attitude towards AI use in class, do I escape the charge of authoritarianism, because I am not technically banning it, merely creating a class culture in which AI use just is not done?
Yes.
Whether it ages well depends on the type of claim we’re assessing. The evaluative claim (fine wine that aged well) may be true: this won’t be a better world, at least autonomous thinking-wise. (Is that sufficient to justify catastrophizing? I don’t know.)
OTOH, the prediction or descriptive claim is under-specified. Most wines don’t age very well. And you can’t always predict which ones will. What are we predicting? That students will continue to increasingly use AI? You bet! That they’ll be better for it as thinkers? We’re in evaluative territory here. I’m inclined to agree with you that probably not, but I don’t want to catastrophize either. Or is the prediction that humanists won’t find ways to foster critical thinking and creativity in an AI-dominated world? I have to say I have not seen enough to confidently say.
For all these reasons, it’s not obvious to me that the catastrophizing will age like fine wine, if only because I’m not sure what claims we’re assessing and what would make them true.
Hear, hear.
In the past 2.5-3 years, I have also had an alarming number of interactions with students that have left me speechless. In some cases, it is just the kind of mindless nonsense that you describe, in others it’s sophisticated multi-layered cheating and lying to cover up AI-use. Increasingly, however, I am coming across students who, when asked to complete what seems to me a completely ordinary academic task, fall to pieces without AI. I mean: near or actual panic attacks, inability to write or speak coherently.
When I’m grading essays, I am generous to people who have English as a second language: Philosophy is hard; thinking and reading it in a non-native language is really hard. However, some of my students don’t seem to fully have a _first_ language. When they arrive, they don’t speak or write well, and therefore can’t think well. Careful reading of difficult texts and writing about them, with coaching, feedback and repetition, was the best way of helping people expand their human capacity, enhancing their lives in a very deep way. I despair of the AI era. The technology is stealing the possibility of becoming one’s best self. Soon AIs will be writing the prompts, writing the papers, and grading them. A snake eating its own tail, meanwhile constricting human capacity till it suffocates and dies. We already have a huge problem with a lack of reading, critical thinking, and curiosity. AI will only make it worse. As a matter of principle, I will never use AI in my teaching, even if there are a few cool ways to do things with AI.
I share most of your concerns, including again AI reliance. I also have some discouraging anecdotes, and of course it’s ludicrous to oppose developing “at least minimal faculties” independent of AI. But banning AI from humanities classrooms is obviously (right?) not going to be the solution, and would likely be counterproductive.
I tend to think that by banning AI use as much as possible in my class I am preparing my students to thrive in the world of AI. They’re going to get plenty of practice without us.
Yes, we should aim to ban it.
I will die on this hill.
Your entire profession is going to die on this hill. Any field that doesn’t successfully incorporate AI is going to be extinct very soon.
I’m not sure exactly what a “field” is, nor am I sure about what it means to “incorporate” AI. I know that I teach philosophy, and I believe in the importance of teaching philosophy because I believe in the ways in which philosophy can distinctively help train people to get their head around the complexities of a serious problem, work through possible answers to a problem, and then articulate and defend an answer. I don’t know–because I don’t think anyone quite knows–exactly how AI is going to fundamentally change the world our students will be facing, though I think we all know *that* it will do so in unforeseeable ways. In the face of all that uncertainty, I hold out hope that it will remain important for people to think through and compellingly answer hard problems. My hope is to do that, and keep AI as much as possible on the sidelines, for as long as I can.
Fine. I have no desire to spend the next thirty years interacting with a chatbot. Less than no desire.
How soon is very soon, Eric? Pick a date by which you’re pretty sure philosophy will be extinct if it doesn’t incorporate AI. I’ll make sure to set a reminder and check back with you.
The message from students is clear: the old way of doing philosophy has been automated. So, these students will go into the world, vote, become administrators and legislators. And slowly but surely, academic philosophy will wither. It probably won’t *entirely* die. So, when I say extinct, I mean the extinction of pretty much the entire philosophical infrastructure we know today. It will be taught rarely, and only at the wealthiest universities.
On to dates. If you’re a provost or college president, what are you going to fund? The Luddites with the blue books, or the folks with the high-tech computers? Not the blue books. There are already plenty of signs of this happening right now. Just read a little about how admins are strongly encouraging the use of AI across the curriculum. For the kind of extinction I’m talking about, there will be some generational change involved. So let’s say 20 to 25 years. I’ll probably be dead by then.
Of course, philosophers might start to adapt. You can read plenty of good articles on how mathematicians and linguists are adapting. And philosophy might come to flourish. I hope it does. But I fear that the Luddites seek to destroy it.
I’d appreciate some citation here; this response certainly is not what I have heard from my own students. Though, I think this is actually irrelevant. Shouldn’t it be professional philosophers who decide what the best norms of our discipline ought to be? Surely professors are better situated than novice undergrads to make an informed judgment.
If politicians will not fund a discipline merely because it is low-tech, that is a very bad reason. As a general principle, philosophers should not warp their practices so as to accommodate politician’s very bad reasons.
To be clear, doing philosophy the “old way,” as you call it–philosophy without AI–is the only way philosophy has been done in the history of humanity. The old-fashioned way is the method by which every current member of the discipline was initiated and trained in philosophy. That’s a very strong and well-attested record. The old way is imperfect, but we know that it can successfully produce philosophers.
In contrast, the problem with generative AI is that we do not even know if it can be used to make philosophers at all, let alone in a way that is superior to the old methods. How, specifically, at the level of assessment, does using AI improve on the old way of training up philosophers? You suggest below that philosophy classes might “be more like workshops. Like in logic classes where the students just work through a series of proofs, or argument extraction cases.” The problem is that there are two ways this might be implemented: (A) You let students use AI to instantly solve the proofs and extract the arguments, or (B) you restrict student usage of AI to preserve their learning. On (A) students learn nothing because they don’t do any way work, and (B) you’ve rejected already as what the “Luddites” are doing. So, you and other proponents of AI in the classroom are going to need to provide a lot more detail on how this is going to work. And until that is done, it is hard to see why caution shouldn’t convince us to stick with educational method which we know has a chance of working.
You’re raising some very deep questions here.
But first an easy point. You wrote: “Shouldn’t it be professional philosophers who decide what the best norms of our discipline ought to be?”
Yeah, maybe it should. But we don’t live in a world whose ideals were crafted by philosophers. And when did we ever decide on these norms? Never. They grew up, organically, without any reflection. And they were mostly shaped by factors entirely outside of philosophy (e.g. efforts to turn philosophy into a science, norms developed as academic productivity became more quantifiable, etc.). So we dropped the ball on that point long ago.
And you wrote: “If politicians will not fund a discipline merely because it is low-tech, that is a very bad reason. . . . philosophers should not warp their practices so as to accommodate politician’s very bad reasons.”
Unless, of course, philosophy departments want to survive, and philosophers want to get paid.
You wrote: “The old-fashioned way is the method by which every current member of the discipline was initiated and trained in philosophy. That’s a very strong and well-attested record.”
What’s well attested? There are plenty of critics who argue that philosophy as practiced today has gone so far into irrelevance and narrowness that it deserves to die. Hopefully to be reborn in some new form.
Look throughout this thread at all the grandiose claims made about the benefits and virtues of doing philosophy via essay writing. And all those claims are backed up by zero evidence. They’re not even testable.
Essay writing is not the only way philosophy has been done in the past. We could indeed go back to Socrates, and do philosophy in dialog with AIs. We could be training AIs. The opportunities are literally endless.
The real problem here is your assumption that the only way to use AI is to “let students use AI to instantly solve the proofs and extract the arguments”. Obviously, that’s self-defeating. So don’t use AI that way.
AI chess players can crush all human chess players. And yet here we are, living in a golden age of chess among humans. Not to mention the fascinating accomplishments of human-AI centaurs.
But I’ll agree that this is an X-risk for philosophy. Either philosophy solves it, or philosophy dies.
The golden age of chess is not one in which the players each enlist AI to beat their opponents. It’s one in which they (1) learn to play chess for themselves, and (2) regularly play chess against one another without being fed the optimal move by someone or something else.
This is not what students are doing with AI. It is, however, what I am teaching them to do. It’s also what I do in my research.
So your assumption, like pretty much every Luddite on this thread, is that the only way to use AI is to use it to do all the work for you.
I play against a human. After the game, I go through my moves with an AI coach, which helps me learn where I made mistakes and where I did well. And the AI coach can see that I have trouble using bishops, or that my opening strategies decline around move 5, and so on. And I become a better human chess player.
Or I pair up with a chess engine and we play as a centaur against another centaur.
Or when I would play against much lower rated players, we could set an AI to play somewhere between my rating and my opponent’s rating, so they could start to see how to play better.
Or, well, there’s dozens of use-cases. But you can’t see them. I’m betting you can’t see the positive use-cases for philosophy either.
That’s not how students are using it.
I’m not assuming that’s the only way to use it. I’m drawing a strong inductive inference that this is the primary way students are using it.
Because the assignments profs are giving them are old-fashioned essay writing assignments. Stop giving those kinds of assignments.
What a thing to assert. You’re making the claim that students are cheating and lying “Because the assignments profs are giving them are old-fashioned essay writing assignments.” In what sense is “because” being used here? These students are knowingly cheating and lying and it sounds like you’re blaming their teachers for it. Students are not just using AI to cheat on “old fashioned” (as if essays were a fashion) assignments. They are using it to cheat on everything they can. Why do you think they would not cheat on whatever AI-integrated assignment you have in mind?
The people in this thread are proposing that we cease “giving those kinds of assignments” any more by having students do their writing in class. But it’s clear this triggers your “luddites” allergy. You don’t just want us to give different assignments; you also insist that those assignments be AI-compatible. So, as I’ve said elsewhere, the onus is on you to show that such an assignment exists. The rest of us have not found a way to keep AI, when it is allowed in part, from replacing student effort wholly. What is an assignment on which students could use AI in the way you want, but could not use AI so much as to avoid learning? If there is no such assignment, then your position is a non-starter because teachers are in the business of helping students learn.
For philosophy departments to survive, it is insufficient that people continue to be paid to do something under the label “philosophy.” If, in order to receive funding, a philosophy department switched to producing propaganda for a totalitarian government, it would cease to be a philosophy department. Philosophers can be guided by pragmatic considerations only so far.
I think this strikes at the heart of things. It sounds like you think AI *will* be very bad for the discipline of philosophy, but that doesn’t matter to you because you think the discipline of philosophy, as it stands, isn’t worth preserving. That’s consistent, but I’m not sure why that position on AI would be attractive to any of the rest of us who *do* value philosophy as it is currently practiced.
You agree that there must be some restrictions on student use of AI; what will you do if students insist that they’d like to use AI without those restrictions? Will you concede, “The message from students is clear: the [new] way of doing philosophy has been automated”? The advice “Don’t use AI that way” is unhelpful. Many of us are working very hard to prevent students from “using AI that way,” with limited success. Your suggestion that students could be tasked to dialog with AI doesn’t engage the problem; AI can be prompted to fabricate a conversation with itself.
Throughout this thread you seem to enjoy (a) calling people Luddites and (b) claiming your opponents make claims without evidence. In response to (a), I would say that you have no evidence for accusing anyone of being a Luddite. Wanting a narrow ban on generative AI in philosophy education no more makes a person a Luddite than wanting a ban on nuclear weapons technology does. We all see that AI has tremendous power–including for the good–in many domains of human life. But we see this particular application as harmful to our students and worth resisting. In response to (b), I would suggest that you consider that AI is itself a very new and untested technology. We might find, as we have with other technologies, that there are massive downsides that we could not see from the outset. That is reason for caution and for demanding data on the effects of this technology *before* mass adoption rather than after. In other words, the burden is on you as well to provide evidence for the helpfulness of this technology that outweighs the risks. That includes providing an explanation of how AI could plausibly be compatible with our students’ learning rather than vaguely claiming, “The opportunities are literally endless.”
All very good points!
All I’m saying is that the old way of doing philosophy is going to disappear, whether we like it or not. So we should try to figure out new ways of using AI to do philosophy better. Mathematicians and scientists are already doing that. Philosophers can too.
Given the almost total hostility to AI on display here, and the astonishing commitment to the old ways, I’m happy to call people Luddites. The term fits.
The opportunities for using AI are endless. Some of those opportunities will succeed, others will fail. Our job is the hard work of testing them.
Imagine a thread where every commenter explains their new way of using AI in philosophy teaching, and evaluating it for pros and cons.
But no, what we get here is “back to blue books”, “ban AI”, etc. The anti-AI crowd (as I say, the Luddites) aren’t even bothering to think about the future. And departments will keep on closing.
Eric, I would urge you to actually read a social history of the Luddite movement before you follow the rest of the uninformed commentariat in using it as a negative pejorative. The Luddites were (in part) reacting to the exploitative and profiteering activity of an unspeakably uncaring capitalist class. Smashing the machines was the only way to alert the world to the fact that entire ways of life were being fed into a meat grinder, with poverty and starvation the predictable result.
I am a proud Luddite. Anyone who would choose to prioritize the smooth functioning of the machines over the human lives they are ruining is not only morally corrupt but an enemy.
Really, read up on it. Even just the wikipedia page would be a good start.
I know the history of the Luddites. The term has many senses, and I am using one of the common contemporary senses.
You seem to think it is an insult. If it means “someone unthinkingly against all technologies”, then it is inaptly applied to the people in this thread that you’ve been calling “Luddites”. If it means “someone against some technology/worried about the effects of some technology” then it is not an insult—nor is it a common use of the term “Luddite”.
Also, calling people pejorative names is rude and unnecessary, and it’s not helping your case.
I don’t think it’s an insult. It’s a descriptive term. And it’s apt here. For instance, https://www.insidehighered.com/views/2023/04/03/case-luddism-against-chatgpt-opinion
Professor Fritts’ article is centered on education, but the same worry (that use of AI threatens our humanity, defined as authentic self expression) applies to philosophical research as well. If students should be banned from using AI in classrooms, then faculty should also be banned from using AI in their research.
Professor Fritts’ argument is an important one, but I am not quite convinced that AI so clearly threatens authentic self expression. How does this happen? What does it mean to authentically express oneself? Professor Fritts must have in mind examples of students copy-pasting entire essays, and in such cases the inauthenticity is clear. But if I use chatgpt to source for examples for an argument, am I being unauthentic? How is this different from say sourcing an example from a friend or colleague?
Professor Fritts acknowledged that AI will pervade our lives regardless of how humanities decide to deal with it. A proposal to ban AI then, must mean the irrelevance of the humanities.
Yes, what is different about asking a computer for something versus asking a friend or colleague for something? I certainly would happily replace all my conversations with friends and colleagues with conversations with a computer if this were possible, and I look forward to a future in which this is indeed possible, as it seems not so very far away. Someday perhaps I can do away with humanity altogether, with respect not just to conversations but to everything!
Sometimes at the very edges of my cognition a worry in the shape of “but what if others think as you do…” impinges on my existence, but it is easy to put this out of mind by asking ChatGPT for some strategies to eliminate intrusive thoughts. I then go back to daydreaming about a paradise without other humans.
You’re referring to conversing with colleagues for pleasure whereas I was talking about conversing with colleagues to fill a practical need (namely finding an example for an argument).
Anyway, is this argument supposed to apply only to AI? If so, why? If not so, then it seems absurd. In the good old days, people might consult their neighbors if they needed help with directions. Nowadays we use google maps. Is this supposed to have led to some sort of alienation or loss of humanity?
I was not referring to conversing with colleagues for pleasure.
Once I had a GPS device tell me to “navigate offroad,” which was a gentle description of what the route in fact entailed, which was driving over a cliff into the ocean. (This was when I was driving some philosophers around during a conference, so perhaps someone reading this remembers the occasion.) I still use GPS to navigate (or, rather, I would if I drove anywhere anymore) but I did not trust it then when it told me to annihilate myself and my passengers and I would not trust it now if it gave bad advice.
The problem is that students are typically not in a position to notice if AI is giving them the intellectual equivalent of “drive off a cliff.” Students who ask AI instead of coming to my office hours for help tend to get things quite wrong. But, insidiously, they do not always get things wrong, because the AI is just bullshitting, and the bullshitter doesn’t care if they get it right, so sometimes they do. So I cannot just say “the AI always gets it wrong.” I can only say “the best you can do is use your judgment, and your judgment is not good enough.” Some listen. Others don’t. The AI is very seductive. Some models (most?) will even flatter you and tell you your judgment is very good.
The way to get students in a position to be able to use AI responsibly is to have them practice these skills without AI. They need a lot of practice. So, we should do our best to keep them from using AI.
> The way to get students in a position to be able to use AI responsibly is to have them practice these skills without AI. They need a lot of practice. So, we should do our best to keep them from using AI.
I don’t think the last step follows, if it is intended as a universal, rather than as a generic. There are certainly some contexts in which we should keep them from using AI, so they can develop skills without AI.
But they should also develop skills of AI use.
Just like we want students to both develop written expression and spoken expression, and develop reading skills and listening skills, we want them to develop skills without AI and skills with them.
There are many differences between asking a friend or colleague something and asking an AI. You have obviously identified some of these differences, noting that there is a kind of personal knowledge and understanding that one gets from asking a friend or colleague that one doesn’t get from asking an AI. But there are many others as well.
Not all of us have friends or colleagues who are experts on every literature. Not all of us have friends or colleagues who answer our e-mails at the hours that we are working.
There are also values to getting second and third opinions from different sources – often the more different the better. AI is usually a bland averaging of a certain sort of conventional wisdom in a field, but there is value in that as well as the recommendations of an opinionated person.
I also note that the form of your reply is completely in keeping with the view of Socrates that we lose our ability to debate and ask questions by learning from books. I agree with Socrates that it is usually best to get an understanding of a view by having an extended conversation with a person who holds the view. But books are nevertheless quite valuable, even though most of us would not want to live in a world of books only without other humans.
Plausibly Socrates was correct about certain things being lost when we make use of books. (Perhaps growing up in an exclusively oral tradition tends to give us increased recollection powers, for instance.) On balance I think books have made things better, though.
AI does not seem to me to be a tool like that. On balance it seems to me AI causes atrophy in skills that are more crucial than those that suffer atrophy when one turns to books. Recollection skills are not very important if the information can be found in books, or, later, through Google. But critical thinking skills are very important even if AI can take over critical thinking, because until AI can be trusted to do all the critical thinking, one must be able to check AI’s answers. And one cannot do this of one’s critical thinking skills have atrophied.
There are two main replies one might give to this. The first is that reliance on any third party can cause the atrophy, so this is not an AI-specific worry. The second is that the atrophy caused by AI reliance is so minimal as to be not worth worrying about, especially given the benefits of AI use.
With respect to the first point, there are lots of natural limits on how much one can rely on humans in ways that tend to prevent atrophy. If you ask too many colleagues too many questions about concepts that are too basic for you to properly be outsourcing your thinking, your reputation will plummet, you’ll get feedback from people that you really ought to be figuring this stuff out yourself, etc. AI has no built-in limits like this and will happily let you reduce yourself to an unthinking robot-regurgitation machine at breakneck speeds. (Consider how we will not simply feed answers to students in office hours.) There are also practical limits: as you yourself point out, colleagues will not answer your e-mails at all hours. Even if they do, they might say “I don’t know, I’ll have to look into it.” AI will answer instantly, always, and even if it doesn’t know the answer it will bullshit to your face. This combination of easy access and disregard for truth makes it much easier for you to turn off your brain when asking AI compared to asking people.
With respect to the second point, this is of course an empirical question. I must say that from my interaction with students I am not impressed with AI’s capacity to promote critical thinking, and indeed it seems to either retard or actively destroy critical thinking capacities. I have less experience interacting with colleagues who make extensive use of AI in their research, in part because it’s too early for there to be many people like this. But, I look forward to the chance to evaluate people like this in the future. Perhaps my worries will turn out to be overblown with respect to people who developed their skills the old-fashioned way before AI existed. I do still think my worries are applicable to students, though, which underscores my very AI-phobic approach to pedagogy.
Unless this is applied across all classrooms, students will be used to other classes or even other universities that integrate LLMs, which has several implications. One is that unless you’re clear about what kinds of thinking you’re developing, students might spend their cognitive load on “not sounding like AI” over thinking in the way you’ve prescribed.
Can a generalization of this coarseness of grain get a useful answer? (He responded without reading ahead.)
People who are thoughtful design classes with all sorts of considerations in mind. Some will have good reasons for banning (even decisive reasons), some will have good (even decisive) reasons not to.
Our students are entering a world in which most people use cars to get around. Cars can also drive more quickly than any human can run. Therefore, when we train them to do complete a 10K run for physical fitness, we should encourage them to drive their cars that distance instead. It gets them there much earlier, avoids stress and sweat, and boosts their self-esteem.
On the other hand, physical fitness is important, and you get a lot of physical fitness by walking or running, so when we visit the Bertrand Russell archive to do a class field trip, we should encourage all the students to walk to Hamilton, Ontario rather than driving or flying.
More seriously, when it comes to the use of bikes vs cars vs planes vs feet, we should try to understand what it is that we want to accomplish with a given bit of physical movement, and see which of these means will be most useful for what we want to accomplish. Similarly, when it comes to a given piece of writing, we should try to understand whether this piece of writing is one that is aiming at developing some of our writing skills (and if so, which skills) or whether it is aiming at clear and concise communication to a particular audience, or something else, and we should use the tools (or lack of tools) that best help us achieve our goals.
My many cents:
Do universities/colleges have an obligation to prepare students with AI-related skills to help them thrive? -Yes.
Does that mean that developing AI-related skills needs to be the learning outcome of every course? -No.
Do students need AI-free skills in order to thrive in a world filled with AI? -Yes.
Do universities/colleges have an obligation to prepare students with AI-free skills to help them thrive? -Yes.
Courses in what fields are relatively suitable for students to develop those relevant AI-free skills? -Well, humanities?
I agree with every step except for the last – I don’t think the humanities are uniquely relevant for developing AI-free skills, and I also don’t think the humanities are uniquely irrelevant for developing AI skills.
There should be humanities classrooms with AI and humanities classrooms without, just as there are math classrooms with calculators and math classrooms without.
As a college professor, I believe we should strongly consider banning the use of generative AI tools in our classrooms—not because I am opposed to innovation, but because the widespread, uncritical adoption of these technologies risks undermining the very goals of higher education.
First and foremost, the integrity of the learning process is at stake. Our mission as educators is not simply to produce assignments or grades; it is to foster critical thinking, independent inquiry, and original expression. When students turn to AI tools to write essays, generate problem sets, or even interpret course material, they shortcut the cognitive and creative processes that are essential to genuine learning. AI may produce plausible answers, but it cannot replace the human struggle to understand, to reason, and to articulate thought.
Moreover, allowing unrestricted AI use raises serious concerns about academic honesty. Plagiarism policies are built on the premise that students’ work should reflect their own understanding and effort. AI-generated content blurs this line, making it increasingly difficult to distinguish between a student’s authentic voice and that of an algorithm. This undermines the value of a college degree and erodes trust in academic assessment.
We must also consider equity. Not all students have equal access to advanced AI tools, nor do they all possess the digital literacy required to use them responsibly. If we allow AI to become a default academic assistant, we risk privileging those with more resources and technological fluency, further widening existing educational gaps.
Finally, there are pedagogical implications. Higher education is not simply about absorbing information; it’s about engaging with ideas in a rigorous, disciplined way. AI tools, while impressive, often generate surface-level responses and can mislead students with incorrect or oversimplified information. Encouraging reliance on these tools runs counter to our commitment to depth, nuance, and scholarly integrity.
For these reasons, I advocate for a cautious and critical stance toward AI in the classroom. Until we have clear frameworks for its ethical and pedagogically sound use, a ban is not just reasonable—it is necessary.
Was this written by AI?
Definitely. Only AI would break out the em-dashes for a blog comment!
Em-dashes and lists (e.g., “to foster critical thinking, independent inquiry, and original expression”; “to understand, to reason, and to articulate thought”; “depth, nuance, and scholarly integrity”; etc.).
I sometimes worry that I’ve too tightly internalized the idea of giving three examples of things in my writing. But maybe not as much as some AIs.
Not Artificial Intelligence themself, but their cousin, Art Smart.
My ideal-theory self wholeheartedly agrees with Megan. My non-ideal-theory self recognizes both its own limits (I find LLMs extremely useful and interesting) and the range of futures still open to us. The range is wide but the AI-free futures are at the very tail end of a long-tailed distribution. Barring civilizational collapse or global anti-AI revolution, there is no world in which both of the following are true:
(1) Students in our classroom are not allowed to use generative AI in any shape or form.
(2) We have enough students enrolled in our classes for our programs to survive.
I don’t think training students to enter the workforce is a good objection to Megan. Maybe some fields can and should evolve to incorporate AI and that’s where students will foster the relevant skills. But that is perfectly consistent with fields (at least parts of the humanities) that are more or less AI-free zones. Indeed I find that combo aesthetically attractive. So it’s not that we, philosophers, must train them to use AI because we, qua philosophers, must prepare our students for the workforce. Rather, this is an enormous coordination problem. Students, administrators, and human beings in general being how they are, can we imagine a future where we, as humanists, will get away with offering low-tech classes with low enrollment for the foreseeable future? I’m very skeptical. This fills me with despair, despite my own non-ideal-theory self’s susceptibility to the seductions of generative AI. But so do many things about this world that I am—we are—not going to change by fiat.
Philosophy education may well be in an adapt-or-die moment. But there’s more than one way to die. It may be that the level of AI-adoption required for there to be something called “philosophy” in universities going forward disallows the primary benefits of a philosophy education. That’s just to say that you could be right about the situation (and I fear that you are) without there therefore being a clear path forward.
I agree. Maybe there are more dignified ways for us to die than giving up completely.
I’ll also note there’s a spectrum of uses of AI and I’m not convinced that nothing short of a wholesale ban can prevent the slippery slope of enhancing students’ abilities to corroding the atrophy of their thinking skills.
I think (and hope) that’s right. But still, there may be some reason to think this particular slope is particularly slippery: the spectrum of uses you describe seems to be expanding faster than our ability to suggest, test, and settle upon norms for their employment. I’m sure that will level out at some point, but I fear we may concede a lot on the way down.
Oh yeah, it’s expanding a lot faster than we give ourselves credit for being able to handle let alone police.
This seems broadly right to me.
All universities need to teach all students how to work effectively without relying on LLMs, because this ability is going to be key to individual and organizational success.
People disagree about how useful LLMs are now and how useful they will be in the future. I’m going to set aside the alleged possibility that LLMs will lead to “artificial general intelligence.” I consider this obvious marketing hype. Here are more realistic scenarios:
I don’t know what the right strategy is for teaching students to function without LLMs. Just telling students not to use “AI” is not effective. (Even though I catch a bunch of students cheating with LLMs on every take-home assignment, and I tell my classes this, it’s not an adequate deterrent.) We may need new methods to teach old skills. Or we may need to bring back some older methods, like blue books, of course, but maybe also oral exams, or maybe a modernized version of medieval disputation.
I agree with Fritts and have been sharing this post widely. I don’t really have much to add to her argument. What I would say to others in the comments is to maintain healthy skepticism about the claim that LLMs are here to stay. We have two pretty recent hype cycles in Big Tech that failed to turn into profitable ventures: blockchain-based assets and virtual reality. They were both the next big thing and both went completely bust. Ed Zitron has been betting on AI as the next tech bubble, and the case he’s putting together is pretty convincing to me. Case in point: Microsoft canceling plans on building massive data centers, as investment in AI has so far failed to make any meaningful economic impact.
So, yes, in a sense, LLMs might be here to stay, in the same way blockchain and VR are here to stay, in that people are still gambling on cryptocurrencies and a handful of hobbyists own and use VR headsets. But nobody is out here urging a thoughtful and responsible approach to preparing students for a world where these things will be unavoidable, where they ignore them at their own professional peril.
(Note: I’m aware of the people here who say they’ve used chatbots as interlocutors for improving papers, but this is irrelevant to my point, which is that the case for LLMs being unavoidable, and therefore an essential skill for future-proofing students’ careers, is very, very weak.)
An important difference between AI chatbots and blockchain or virtual reality is that the general public never really started using blockchain or virtual reality.
A better example would probably be voice assistants like Siri or Alexa – people started adopting them, and they looked like they were going to be significant, but they seem to have fizzled.
Voice assistants are a good example. Alexa was meant to be way more than a voice-activated stereo.
Worth noting that AI could both be the next bubble and be medium-term transformative. The internet was a tech bubble! The recession in 2001 was mostly about over-investment: everyone thought the internet would be economically transformative, no-one wanted to miss out, and it wasn’t yet clear what the economic opportunities would be. There was a market correction and lots of people lost money. But of course they were *right* that the internet was going to be transformative.
Well done, Megan Fritts, for writing the best piece on the consequences of AI that has yet come out.
Anyone who opposes at all the zombie-fication of the youth (and ourselves) should oppose the widespread use of AI. An increasing number of students spend all day on screens, on headphones, passively consuming short-form videos and responding to any demand on their time by offloading any task onto Chat. Have to write an email? Ask Chat to write it for you. Read a book? Ask Chat to give you the highlights. Write a paper? Ask Chat to write it, then ask it how to get around AI detectors. Wish your mother a happy mother’s day? Ask Chat how to do it. The resistance to any kind of actual cognitive activity is astonishing. This is a growing minority of students, sure … but it’s growing, and I would say a majority of students are well on their way to this state.
I’ve ceased to be surprised by this, but what is surprising is that PHILOSOPHY professors would embrace these changes or react in the ways that we see in some of these comments. Yes, LLMs aren’t going anywhere. But THOUGHT is on its way out, and if we don’t radically change the way we operate in the classroom, we’re part of the problem.
Out of the 27 commenters I counted on this page, I think at most 6 can be read as embracing these changes, and even then only cautiously so. I am one of them. But most readers seem to be missing the target. The question is about the absolute ban of any AI tools from the classroom (and presumably from the homework toolbox, too). This is what some of us are reacting to. Of course, at least speaking for myself, unrestricted use of AI would be a total disaster. That’s not really in dispute.
Since I have reason to believe you were referring to me, among other commenters, I will note, as I often do, that I appreciate knowing whom I’m talking to. I request that my students submit their own work, under their own name. If you are going to make drive-by comments about your colleagues, have the courage to let them know who you are.
There are relevant differences between drive-by comments on a blog and your student assignments (one example: you have no authority over commenters here, who I assume are typically fellow PhD-holding philosophers and professors rather than your students). No, the commenter doesn’t need to show courage by deanonymizing such an innocuous comment, ffs.
Of course, there are relevant differences, it’s just a limited analogy. A relevant similarity is that Prof L’s comment is quite directly, albeit vaguely, berating identifiable fellow philosophers for being “part of the problem” (“the zombie-fication of the youth (and ourselves)”, I take it) while commenting pseudonymously on the internet. This is poor form and does not exemplify the courage to think that I hope we are all trying to inculcate in our students…
Yeah, I overreacted because I was (at the time) annoyed by our earlier comment chain. My fault, point taken, carry on.
No worries, same here. One thing we may agree on: social media has irreversibly corroded human beings’ ability to communicate normally.
Definitely with you there.
The main comment I had in mind was actually by “James”, who is not identifiable.
I just take it that our job is to teach students to think, not to think along with AI, or to use AI well … They will use AI well whether we teach them to or not. (I’m also reminded of advocates of tech in schools who say things like, “We need to teach the children how to use this new technology!” … as if any child really grows up without being able to use a tablet, or phone, or chromebook. The problem is that they don’t know how to do anything without those things). So, partly for those reasons, I’m absolutely in favor of a total ban on AI.
I have my own reasons for remaining anonymous—courage to post under one’s own name for some might be foolhardiness for others—although in this particular thread I would have been happy to post under my own name … too late now. But, if it’s any comfort, I’m no one of significance, just a middle-aged, middling professor you’ve never heard of at a middling university.
Thanks for your reply, I understand. We don’t know if James is a philosophy professor, though, so my paranoia narrowed down the scope of your criticism in my mind.
I fully appreciate people’s reasons to remain anonymous—I do not systematically post under my real name in every forum. But it does create an asymmetry of reputational costs.
I’ve tried to implement total bans on laptops and phones in the classroom and it just doesn’t work. Maybe I’m just not lucky or terrible at playing cop. I’ve not considered a total ban on AI in part because our administration is encouraging students to use AI, also because they *will* use it, whether I like it or not, and they’re more likely to use it well if they know I condone certain restricted uses of it. It builds mutual trust and, I hope, transparency. But maybe I’ll give the total ban a try someday.
What part of what I wrote was objectionable? Do we not have a duty to prepare our students to thrive in an AI future? Is there no such thing as competent use of AI? Should we never allow a conversation with AI in class??
My guess is that you will agree that yes, we do have such a duty and that AI-independent critical thinking is part of that preparation. And you’ll agree that AI can be used competently. And I suspect that you don’t even think we should never allow an AI conversation in class. (Note that by banning AI from the classroom, you cannot even demonstrate its flaws and limitations.)
Am I wrong?
I think thriving in an AI future means eschewing the personal use of AI for all but the most automatic of tasks–tasks that do not require thought (e.g., asking Chat to comb an essay for instances of some word, repeating those instances back to me in context, ordering lists, scanning a document for possible typos, etc.). I try to make my classroom a tech-free place — I take it my duty as a professor is to allow them to exercise those capacities that they have receded from their education. They can think, read, write essays without any technological assistance, apart from pen and paper. Am I constantly battling with students and their phones/headphones/tablets? Do some gripe about it? Do I have to make exceptions for students with accommodations? Yes, to all these questions. Is it worth it? For sure.
Obviously I think we should never allow an AI conversation in class. That’s directly implied by “total ban on AI in the classroom”. Students who utilize this technology frequently will be far more sophisticated users of it than I will ever be, they have increasingly grown up with it; it’s infiltrated their lives, mediates their personal and professional relationships, and some students have outsourced all their thinking to this machine. What could I possibly show them that they don’t already know? Students who eschew use of LLMs—well, I think they’re the future, and I’m not going to force them to use it.
This is a great comment, and echoes a lot of what I think.
It seems to me that there is a lot of loose talk about “using LLMs well” that trades on an ambiguity between using them competently and using them responsibly. When anyone says “we have a duty to train our students to use AI well” (I understand that that’s subtly different from what James said above, but still this captures what is I think a common sentiment among cautiously pro-AI profs) part of me wants to scream. We have a duty to train them to use AI responsibly, I suppose, which given the utterly addictive nature of the thing means mostly, in my opinion, a duty to train them in practices of resistance to its temptations.
We definitely do not have a duty to train them to use AI competently. As you say, what could we train them in that they don’t already know better than us? And, if we have objections to AI use, making them more competent users is likely to make them more invested (and thus more corrupted) users.
Jordan, of course I believe AI use should be responsible. (And I didn’t mean to make anyone want to scream.)
Yes, students should certainly be taught tech-free skills. And no, I wasn’t advocating “forcing” students to use AI.
But what I don’t get is that you want to impose an absolute ban on something you admit you don’t understand as well as your students.
If you’ll humor me, have you experimented with any of the more advanced AI systems, or has your experience with AI been just lazy students relying on AI slop?
N=1, but before I took drastic measures, the “lazy students relying on AI slop” were literally 30-32/35 in all 8-11 of my sections. My colleagues reported similar levels of AI use, which is why we shifted most of our assessments to invigilated in-person work.
It’s just not a mere handful of them. It’s the overwhelming majority, and it’s followed by so. Much. Lying. Including for the lowest-possible evaluative stakes. You cannot teach the necessary underlying skills in an environment like that.
There is an interesting analogy with tools like Mathematic and Matlab in physics and engineering. For the last twenty years or so, we have had tools that transformatively improve one’s ability to do calculations and simulation. To the best of my knowledge, everyone in those fields is in favor of students learning how to use them and integrating them into their workflow. But equally: to the best of my knowledge, everyone in those fields thinks students need to learn the core ideas of algebraic manipulation, integration, etc, and not just mindlessly input formulae, because those tools are only useful to people with a solid understanding of the concepts they rely on. And that means examining some fraction of student work in a tech-free, ‘blue book’ space. (Arguably you could say something similar about spellcheck.)
It is perfectly possible to imagine a similarly conciliatory resolution of these AI issues, where everyone accepts that core aspects of writing and structuring an argument need to be understood by students and need to be taught and examined in an AI-free space, but where everyone also accepts that people who understand these core ideas are then going to use AI as a major force-multiplier.
I don’t know if this is actually desirable, though: part of the problem is that the technology is moving so rapidly that we don’t yet have a clear understanding of how helpful AI will be even to people who do understand the core ideas of writing and argument. (I haven’t yet found it of more than peripheral use in my own academic work, though I’ve not spent that long trying to learn.)
That’s a good analogy. And I suspect that the philosophy classes that teach the skills you mention aren’t going to be read-lecture-essay courses. They’ll be more like workshops. Like in logic classes where the students just work through a series of proofs, or argument extraction cases. Such classes really will aim to teach well-defined skills. Contrast with the Luddites who insist (without evidence) that students learn nebulous virtues by doing philosophy the old-fashioned way.
I’ve used Matlab quite a bit, but I never felt like it was thinking for me … It was just a really advanced calculator on steroids (with very pretty outputs). I’m not sure how to say what I think the disanalogy is here, but it has something to do with human creativity.
Another thing about AI, as it creeps into the cultural and personal, is that it is extremely “conservative” in the sense that it reinforces all the norms and ideals and ways of thinking/speaking that are in vogue in the current moment. Ask it a personally/politically sensitive question … There’s not a sense there of the diversity of viewpoints that might (or perhaps more importantly, could) exist. I do think people turn to it as a guide for their behavior in sensitive situations … It’s culturally stagnating. It attenuates personhood. Of course it can be useful in certain contexts, but … IDK. Matlab never took over my personal life (maybe that’s why I will never be a physicist, HAHHAHAH), but AI could.
‘Preserving…. philosophy’
This says so much in my view. I think it admits of a view of philosophy as essentially dead, but of value and so needing to be preserved for the good of some. Like keeping your grandparent alive on a respirator because you want them to meet their great grandchildren. Or preserving an old gas powered car because there is value in having it around in a world of electric cars.
I reject this notion. I think we need to have less respect for the standards and traditions of the field and profession and more for the value of philosophizing. Philosophy is perfectly capable of transforming into something better equipped for the new world (better equipped even than the world itself) but we all love philosophy the way it is far too much to create something more.