BS Universities: The Future of Automated Education

Speaker 1 I think behind all this is a very big question about what the university is for. If it's just instrumental, if it's just about getting students in and out with the degree, well, then AI. But, of course, as Rob's already indicated, that's self defeating because you're gonna have graduates who don't know how to do anything. Right? They're gonna go into law or engineering or into medicine, and they won't be recognized by professionals in those disciplines as actually knowing their stuff, as knowing how to do it themselves. So even that instrumental model and its sort of attraction to generative AI or to AI systems seems to be self defeating, even aside from the fact that I don't think it's the right way to think about what university education or education in general is and should be about.

Speaker 2 Hello, and welcome to episode 175 of the Human Restoration Project podcast. My name is Nick Hummington. Before we get started, I wanted to let you know that this episode is brought to you by our supporters. Three of whom are Corinne Greenblatt, Kevin Gannon, and Simeon Frang. Thank you all so much for your ongoing support. We're proud to have hosted hundreds of hours of incredible ad free conversations over the years. If you haven't yet, consider rating our podcast in your app to help us reach more listeners. And of course, you can learn more about Human Restoration Project on our website, humanrestorationproject.org, and connect with us everywhere on social media. Any assessment of the potential of AI to contribute to education must begin with an accurate understanding of the nature of the outputs of AI, my guests today write. The most important reason to resist the use of AI in universities is that its outputs are fundamentally bullshit. Indeed, strictly speaking, they are meaningless bullshit. That particular term of art may appear to be attention seeking or dismissive of the issue of AI entirely, but it's actually the root of a much deeper philosophical critique, like the late anthropologist David Graeber's notion of bullshit jobs, but leveled at generative AI in the way it distorts the purpose and function of teaching, learning, and even of education itself. My guests today are Robert Sparrow and Jean Flenity, professor and lecturer, respectively, in philosophy at Monash University in Melbourne, Australia, where they join me from today. And they are collaborators on two recent articles, bullshit universities, the future of automated education, and cut the bullshit, why generative AI systems are neither collaborators nor tutors. As a heads up, if you haven't guessed yet, we're gonna be saying bullshit a lot. Sometimes in an academic context, sometimes not so much. Sometimes it'll be hard to tell the difference. Anyway, I hope you enjoy this episode and learn as much from it as I did. Gene and Robert, thank you so much for joining me today. Good day, Nick. Hi, Nick. Thanks very much for having us.

Speaker 3 So I'm professor Rob Sparrow. I'm a professor of philosophy at Monash University, and I work on science and technology ethics.

Speaker 1 I'm doctor Jane Flannerty. I'm a lecturer at Monash University, and I initially worked on German idealism, so the history of philosophy. But more and more, I'm interested in the kind of implications of new technology on the exercise of what I call human rational urgency, so the impact of new technologies on our capacity to be free.

Speaker 2 Well, we've heard environmental critiques of AI, historical critiques rooted in past responses to automation like the Luddites, sociological critiques, but yours is the first formal philosophical critique that I've encountered that begins with epistemology and goes from there. As I mentioned in the intro, the use of the term bullshit might seem casual or crass, but it has its origins in Harry Frankfurt's On Bullshit, which I remember reading as an undergrad when it came out, I think in 2005. So what led you both or you, Rob, to connect Frankfurt's idea of bullshit to this new supposed age of gen AI? I think one of the first things that strikes people when they engage with AI is the weirdness of this these systems that produce senses, but clearly don't mean what they say. There's nothing behind it. They, you know, they'll flick positions

Speaker 3 regularly, and there's no way that you could check whether they mean what they say. You know, if you if we're talking to another human being, you get a sense from their tone, from the way they behave, whether they're serious, you get none of those with these systems. And and so the idea that there was something problematic about the status of these claims, I think, becomes apparent quite early. And then when you learn that they're trained by getting people to click the little thumbs up button. They're really desperate for our attention, and they want us to behave in certain ways, to believe what they say, to upvote their claims. And so you put that together, doesn't mean what it says, and it's just trying to get the audience to respond in a certain fashion. Well, that is classic you know, that is Frankfurt's definition of of bullshit. And so we wrote a blog post about that, and then we began thinking about, I guess, what it might mean for the status of their claims in other contexts where people rely upon them. And then we started to see universities around the world encouraging their students to use AI or setting, you know, writing tasks where you critique the outputs of AI and thought there's something really wrong with putting this bullshit at the heart of the educational project, and and that's what led us to write this paper, bullshit universities.

Speaker 2 It really brings a whole do different approach to that, that old idiom. Right? Like, bullshit in, bullshit out in programming or coding or information systems. Right? Like, it literally sort of becomes this vicious cycle of, of bullshit. I don't know how else to say it now. It's it's trapped in my in my brain forever. Gene, you had mentioned your interest in ethics and technology. How did you and Rob end up connecting on this project? Well, I think in the hallway at Monash University,

Speaker 1 experiencing a kind of dystopian set of fears about the You guys had a meet cute? Is that that what ends happening? Well, I just You dropped your books and went to pick them up. Absolutely. Loved eyes for a moment. So the the the worry is that universities without really and this is not Monash in particular, but universities in in general, are very eager to adopt this technology without a thorough analysis of its potentials and its and its limitations. Now a lot of the higher education literature will talk about a responsible kind of rollout of AI without, wanting to kinda tackle either the deep seated philosophical issues that Rob and I are interested in, you know, what is the content of these outputs? But also not wanting to look at it in a lot of there are exceptions in the higher education literature, but not wanting to look at it more systemically in the context of of the university now as a whole, right, as a as a corporate enterprise, as a as a profit driven enterprise. Right? So that's really where our contribution that's our intervention, really, in a lot of ways.

Speaker 3 Nick, if I can just come back to something you said earlier about bullshit in, bullshit out. We do think all the outputs are bullshit, but it's important to understand that this isn't because we think that necessarily the inputs are bullshit. It actually wouldn't matter if these things were trained on really good data. I mean, we know that, actually the data on which these things are trained has all sorts of bias in it. You know, it's it's racist and sexist because the Internet tends to have a lot of racism and sexism on it. There's all sorts of problems associated with the data that these systems are trained on, but that's not our critique. Our critique is not that these systems are inaccurate. It's not that they're biased. It's that they don't mean what they say, and they're just trying to in fact, they're entirely oriented towards generating a certain response from us. So they're not oriented towards the truth in the way that you would expect your lecturer to be to be oriented. So this is not about them being inaccurate or biased. This is a deeper problem. They are simply not capable of meaning what they say because of the kinds of entities that they are, which is that they don't have bodies, and they don't share a world with us, and they can't be held responsible for what they say. So the problem, yes, is not just that there's bullshit in, and there's a lot of that, but it's that they simply can't mean what they say. And because they don't mean what they say, we think there's a sense which they don't actually they don't mean anything at all.

Speaker 1 I think it's really important because every time we either of us or both of us give anything like this talk, there's always questions like, but the systems are getting better. What if they're trained on appropriate data? This is a critique about their fundamental constitution, and that it's really, really important to push back on the kind of criticisms that we get.

Speaker 2 And that critique then is that these systems are simply they may be capable of being truthful, but that misses the point, is what I hear, Rob, you're saying there. The point is that they are simply designed to provoke a response from the listener in the same way that a bullshitter, these machines not inhabiting the same space or the same world don't respond to the same, I guess, set of, I suppose, incentives, social cues, anything else. I wonder that what what in your mind is is lost as universities lean into these tools and the technologies, perhaps from the lens of teaching and learning, perhaps from a, you know, a systems lens as educators yourselves at university? What's your take on that?

Speaker 3 So again, I I I wanna clarify that there's we would resist even the claim that they can be truthful. They can the instance that they output would be often would be true if a human being said them. But when but when the machines make the claim, well well, they appear to be making the claim, they're not really telling us anything, and so they're not telling us the truth. And you can see that they're not telling us anything because when if you were to tell me something, I would be able to hold you responsible for that. We would be in a moral relationship, and it would be connected to action in certain ways. If you told me, for instance, that you were going to edit this podcast to make us sound more clever, and then in fact, you know, we listen to ourselves online and we sound, you know, we sound like Gummies, then I would hold you responsible for that. And I I would be able to say you weren't telling the truth because I could also tell when you were telling the truth by your actions and the other things that you say. But the machines can't do any of that. They're they're not capable of saying true things or or giving testimony, as we say in in various places, because they're not moral agents, and they're not capable of acting in the world as other human beings, as human beings are. Now putting something like that, which isn't even capable of the truth, of being truthful, even though it might be accurate in a sense. I mean, the claim here is not that we can't read their sentences and assess them by essentially imagining they were said to us by by a human being, and we might say, well, actually, that's quite a good good and helpful response. It would be a good and helpful response if someone if a human being made that, but it's kind of nothing if the machine makes it. The problem with putting those things in education is in some sense, education and educators really need to be oriented towards the truth. That's that sits particularly in the university. That idea of an orientation towards the truth sits at the heart of the classical university project. Universities are communities of scholars trying to learn about the truth. And if you replace teachers with these things that don't care about the truth, aren't capable of being truthful, aren't oriented towards the truth, then that is a really I mean, it kinda drives the stake through the heart of the university as it were. It turns the university into a space where people are well, where the institution no longer cares about the truth, but only cares about getting students to think and behave in certain ways, and that's a very different project.

Speaker 1 So that's a great kind of segue to the sort of impact on pedagogy, teaching and learning at the coalface, so to speak. Because if you think about if I think about the relationships that I have with my students so I try to teach in a in a non hierarchical way. So I wrote my master's thesis on a French philosopher of education, Jacques Ranciere. And what we're doing in the classroom is holding each other accountable for our claims. Right? My students hold me accountable for what I have to say. And through their verbal contributions and through their written assessment, I'm holding them accountable for their claims. Right? You know, in Philosophy of War, it's talking about reason and evidence. Right? That's how you hold someone accountable for what they say. Right? Can you back that up? Right? And in turn, my students ask me whether or not I can back up the the claims I'm asserting. Right? So it's a it's essentially a moral relationship, a relationship between moral agents, right, who have a stake in the relationship that they share with one another, who care about how the other party sees them. And AI just cannot participate in that relationship whatsoever. And we're kidding ourselves if we think that they can.

Speaker 3 If you think about the classic philosophical project, the the the sort of if you think about the the definition of philosophy of as the love of wisdom and the role played by Socrates in a kind of self conception of philosophy, one of the things that is immediately obvious is that philosophy is supposed to be connected to the world and to life. It's not just supposed to be about making clever arguments. You're supposed to live this stuff. Now, you know, it's not obvious that academic philosophers always do live it, but at least they could. And so, for instance, if someone says if someone makes an argument that, I don't know, meat is murder, for instance, as the Australian philosopher you know, Peter Singer says we should care about animal suffering just as we care about human suffering. And so they make this argument to quite a radical conclusion, and then they just go about their lives without changing the way they live at all, you would say, well, they didn't really mean what they were saying. And and indeed, you would say that wasn't philosophy. That was sophistry is which, you know, which is philosophy is classically design classically counterposed to sophistry. So people who argue just who are just engaged in a rhetorical project of convincing their audience for money or for for fame or for pleasure or whatever it is. And and so, again, you can see here that these machines are are sophists or or sophists, that they don't they're not living a life. They don't have lives, and so they can't mean anything that they say. And so in that exchange that Jean is describing, if what the student presents is is the work of the machine, there's no way of holding them to account, or indeed, the student is vulnerable to being held to account for something that they themselves didn't say or or mean.

Speaker 2 Is it the sense then, Rob, that inserting AI into this human process, both of the learning and thinking, but also into a learning community where we're supposed to engage both in the space and person to person, and this is the way that we develop ideas. Is it the sense then that AI poisons that endeavor? In the sense then that it makes people not responsible for the ideas that they purport to either hold or to share or to think or to write about. And teachers, educators then giving feedback on student writing that's clearly generated by AI, and then sitting in a seminar classroom, students parroting lines of bull is there a sense that it pervades the the whole endeavor and just poisons it from the ground up?

Speaker 3 Absolutely. I mean, I I I do think this stuff is deeply corrosive of education, because it's not about that moral relationship that Jean was describing, and it's also not about modeling what it means to be something to care about, the topic about which you're talking. So if you think about your own life and education, most people can remember a teacher who inspired them, you know, or or someone who you were impressed by their gravitas and their their sort of intellectual commitment, and that person made an impression on you. That's not gonna happen with an AI system. People can sometimes be inspired by reading books, but books have authors and people who stand behind the words. There's all sorts of aspects of the interpersonal relationships involved in education that these things can't engage in. They can't offer students the experience of being taken seriously and respected. You know, there's no possibility of that spark where where the teacher says to the student, wow. That's a really good point, and I can see you've really thought about thought about this. And the student feels validated and gains a sense of themself as a source of claims, you know, as a person with a capacity to make a contribution to the world. You know, we are looking moving towards a future where machines mark the work of machines. The students submit papers written using ChatGPT or some other large language model, and then the the teachers run that through grading software, and at no point in that experience is anyone learning anything, teaching anything, or experiencing themselves in dialogue in conversation of the sort that we desperately need at this particular political moment. We need people to take each other seriously, and and that scenario I'm describing where machines mark the work of machines is a kind of nightmare future when no one is taking each other seriously.

Speaker 1 Just on the the idea of of being recognized by a a teacher and you recognizing another teacher, that but it goes very deep because we're we're finite creatures. Right? We have a limited amount of time. We don't actually know how much time we have. And so to choose to invest that time in a discipline and in students is meaningful. Right? Because we have a finite resource of of our life, right, to to invest in a discipline and others. Whereas these machines, even though, of course, they have incredible, horrific environmental impacts, right, they're not able to their investment is is never has a ground zero in the sense that it's not someone choosing out of all the things that they could do in their life, in their short life, choosing to invest and spend all their time sound philosophy or history or what have you, and choosing out of all the people that I could be speaking to right now to be speaking to this student and putting their interests first. Right? That relationship is just it just cannot be emulated, right, by AI. So, again, it goes it goes very deep. It goes to the fact that we're we're living beings, finite living beings. Right? And it's on account of that that we can care about each other. And I just cannot participate in those in those relationships.

Speaker 2 I wanna come back, Rob, to something that you had mentioned about, you know, plugging the AI generated essay response into the grading software that also will be run by AI. I've heard from every level, k 12 in The United States through to university, how more and more of these tech tools are being taken over by AI features within those tools, but then also new and different and varied forms that are intended as these labor saving devices. You know, they're intended to be teachers aids and supporters. I wonder if you could speak a little bit more to how that could become a dangerous, slippery slope of of some kind. And then also perhaps how you see AI bullshit as tied up into the entire ed tech industry at this point?

Speaker 3 The way this stuff gets a foot in the door, as it were, is by companies saying this is going to help you. This this software, this will this will take away the kind of grunt work, the stuff that you don't like, and free you up to do the stuff that you really like and value and that is more important to the students. In part, that's because no one sells software by saying, you know, this will put you out of a job and and render your workplace a dystopian nightmare. They always say things are gonna be better if you buy buy my software package. But that model of the machine as a helper has some well known problems historically, particularly if people are relying on the idea that that the human being is going to be in charge, and they're going and that the the the teacher is going to be responsible for what those systems output. So, you know, at the moment, the idea is I'm going to run the student's essay through grading software, but then I'm gonna check it. And I'm gonna check that the the feedback is appropriate and that the mark is appropriate. And and one thing we know about, you know, human computer interaction is people are really bad at maintaining attention on tasks where they only need to actually do something very intermittently. And, you know, you can pay people, highly trained people like, you know, the pilots of the aircraft can be paid to supervise the auto autopilot. But you and I, if the thing works 95% of the time, we basically tune out. And we check the first couple of SOs, it gets it right, and then we stop paying attention. So holding the the staff member responsible for the those rare cases where the machine gets it gets it wrong, that's that's kind of actually disingenuous when it comes we know that people can't do what they're being expected to do. In the same way, we know students are not going to be able to check the outputs of the AI. This is something that that Gene emphasized in the cut the bullshit paper is people say, look, the students can use well, let them use AI, but they have to check its results, and they have to endorse its results. But one, we know that people in general are very bad at that. But, two, particularly a student who is an expert in the discipline, how are they going to be able to tell whether the machine is getting it right? So they're very these these systems are very dangerous in that context. Not to mention that even though grading is something that academics, you know, typically don't enjoy doing, often don't enjoy doing, it is really important because of that dynamic of taking someone seriously. And and we know, for instance, how students respond to tasks that aren't assessed. You know, if you suggest that your students all should write a blog post, you know, post some you know, a social media post, put it up online somewhere, and then it turns out that that work isn't actually being read by anyone, Students typically and rightly respond very badly to that. If if my students know I'm never going to read anything that they say because I'm not gonna grade it, then they are not going to engage in that process. So I do think it is a kind of poison poison pill there. There's also a real problem here in terms of how students can learn the skills that are necessary to use these technologies well. And I I think there's a generational issue here because I'm you know, as someone who's been writing all my life, I can actually prompt an AI, and it spits out some senses. And if I'm paying attention, I can think, is that what I meant? And then I can assess the outputs of the AI. But that's because I know already what I mean. But I think it's often the process whereby we learn what we mean. So if you just go to one of these systems without having thought it through and you don't think, well, what would I say? If you don't formulate your own opinion and think seriously about the matter, but you prompt first as it were, you simply have no way of evaluating. Is that what I meant? I don't know. Sounds pretty good. Click the send button. But that's kind of teaching our students now. Prompting is not developing your own critical skills. It's not developing the capacities to evaluate the claim. So I can prompt all prompt these systems all I like, but unless I've actually been trained in the discipline that I'm studying, and in particularly in particular, in writing, I've actually got no way of evaluating the out the outputs of these these systems. So we are depriving students of the opportunity to actually learn how to think, indeed, to learn what they think when we encourage them to use these systems. You mentioned the ed tech industry, and there is a long history of, you know, to be frank, educational bullshit. I mean, there is a kind of education is really hard. It has it has to be said. Like, the you know, people are very different. The context is very different. You know, something that works well in classroom one classroom fails dismal dismally in another. People respond to changes, so something that works the first time might work the second time. This is this is genuinely to be a good teacher is a kind of lifelong project. People love the idea that there's some tech solution. There's gonna be some sort of magical fix. You know, we'll give the students all computers, and somehow, because they're writing on a computer, that will mean that they're doing something different to what they were doing when they were writing with a pen and paper. So there is a long history of snake oil in particularly educational technology. I mean, depressing, think, actually, the history of pedagogy is marked by or pedagogical pedagogical theory is marked by sort of fads and quackery as well. And, you know, some people have clearly, you know, care deeply and well intentioned, but there are clearly fads in education that we look back on with regret. I suspect this is going to be one of them. I think that, actually, pretty quickly, universities will work out that the way we're heading at the moment will teach our students almost nothing, and that the reputation of the university and the student experience as well will crash badly, and then people will actually go back to a model where they put students in a you know, small groups in a classroom with someone, a human being who's passionate about the material, and they, you know, they write and they read together and they talk to each other, and the technologies are, you know, kept in their place, which is often actually not not in the teaching environment at all.

Speaker 1 So just to add to that, I think that's all I agree with all of that. It really it's competing models of what an education is for. Right? So in Australia, we had something called pushed through by a conservative government. We had the job ready graduates package, which was a sort of major reform, not one that I agree with, of the university sector. And it was a it was a concerted and explicit attempt to present higher education as instrumental. You go to higher education to get a job. What we want is job ready graduates. And if that's how you think about higher education, then of course, Gen AI. Because what what are you there for? To to get the grade, to get the piece of paper, and get out. But if you believe in the humanistic tradition that we're trying to educate citizens, people who can mean what they say and stand behind it with reasons and evidence, who can look other people in the eye and say, metaphorically or otherwise, this is the hill I'm gonna die on because I believe this to be true. Right? AI does does not support that. Students don't even get a start in writing out the sentence. I will argue that or I think that because what they think is the result of some question that they've prompted and that I think is spat out by a machine that they then don't have the expertise to genuinely check or interrogate. So that's you know, I think behind all this is a very big question about what the university university is is for. For. And if it's just instrumental, if it's just about getting students in and out to degree, well, then AI. But, of course, as Rob's already indicated, that's self defeating because you're gonna have graduates who don't know how to do anything. Right? They're gonna go into law or engineering or into medicine, and they won't be recognized by professionals in those disciplines as actually knowing their stuff, as knowing how to do it themselves. Right? So even that instrumental model and its sort of attraction to generative AI or to AI systems seems to be self defeating, even aside from the fact that I don't think it's the right way to think about what university education or education in general is and should be about.

Speaker 3 If I could pick up on the the idea of skill that Jean mentioned, Nick, because that is something that it's very clear that, you know, be becoming, learning, being an educated person, the process of education is not just about knowing stuff. It's not just about a list of facts, but it's actually about cultivating a certain set of skills, developing a certain set of skills, and indeed, I think becoming a certain sort of person, you know, again, picking up on that idea of a humanistic education. So, I mean, it's pretty clear that AI can't teach people certain practical skills, that there are things like, you know, practicing surgery, for instance, that you are going to that require you to, you know, feel a kind of tension in the flesh and, you know, deal with the fluids or whatever. That's not something that one can, you know, just prompt AI. How do I, you know you know, okay, Google. How do I remove the appendix? That that doesn't get you. That doesn't how to remove the appendix.

Speaker 2 Here's what you need to know. Why is an appendectomy before? Alright. Get out your scalpel, Rob. Let's let's go. We just got a live demonstration there. Oh, my goodness.

Speaker 1 So we're about to award you apply a medical degree from there you go. You've graduated.

Speaker 3 And, you know, it's a kind of dramatic illustration of the general rule that the machines are always listening Mhmm. At the moment. Anyway, I am not going to be able to perform the removal of an appendix as a result of that little little spiel. So there's clearly some people will still have to go to university, and they'll have to learn practical things. They will have to do stuff. But we think and in fact, I think it's clearly true that even thinking is a skill. It's something that one needs to practice, and that means and and you're not really doing that when you prompt, or or at least that's quite that's only part of the skill. So we we really worry that students will not learn how to think and particularly how to write because that process of checking yourself and thinking, is that what I mean? No. That's not quite right. Cross out the words and start again. The relationship between the claims is is is wrong here. That is something that you do through writing yourself. And if students stop writing, and there is a real danger that they will stop writing because they have machines that superficially appear to do it better than them, then they really won't learn how to write and therefore how to think. So so we really worry about the impact of this technology on the development of key intellectual skills that students will need in order to be able to achieve mastery of their subject area and be able to do any of this stuff when they go out into the broader world, which might include their their workplace.

Speaker 2 And hopefully not the operating room as we've as we've gotten the demonstration for. I'm thinking here that these technologies, both their adoption and then their use by students don't happen in a vacuum. As as you've both been talking about, the instrumentalization of education means that you're you're there for the most efficient way to get through a degree with the lowest amount of debt perhaps at the end of your, you know, your college education, and to then hopefully get into a career where where you don't have a whole lot of responsibility as we've as we've learned because you might not be thinking or learning a whole lot along the way, but you've got their credential that says that you're there. That that's a troubling future, but clearly one that's being responsive to, right, a certain set of incentives or a certain structure in the system. And I I wonder if you all have come up with some ways to mitigate or resist this idea of bullshit universities in the future of automated education. What would, perhaps, the one hand, universities or policymakers at the big level need to head that off, and what can the little guy do? What can individual educators do in their classroom to help, you know, resist the resist the bullshit?

Speaker 3 In some ways, I don't think that there's any great mystery here. People need to think about the purpose of education and the relationship that between teacher and student that sits, you know, at the heart of the classroom, and they need to think about the impact of these technologies on those things, and they should actually work out pretty quickly that this stuff is corrosive and dangerous. You know, unfortunately, in Australia, at least, we've moved to a world in which the senior management of universities and the governing councils, actually, people often aren't educators or or teachers, and so they seem to they seem to struggle with that. But but at one level, I don't think there's any great the danger to education posed by these systems actually think is pretty obvious if you take seriously the what we know about how people have used automation in the past, how they relate to computers, and you just actually, if you talk to teachers in the classroom and you you talk to people who are actually trying to mark work that is now, you know, a lot of it is generated by AI, and you talk to students and ask them how would you feel if nobody was reading your work or if you were being lectured by an AI, and there, you know, there are people moving towards the idea of having a kind of personal digital tutor would be animated by the same technology that people use to animate do animations in video games, and it will talk to you and, you know, supposedly guide you on your educational journey. Actually, don't think that most students look forward to that future and certainly are not willing to pay the you know, to accumulate the student debt that we expect of our students now for the sake of that experience. So, you know, actually just going out and talking to staff and students would be a start. In a way, I think this stuff will there'll be a nasty reckoning when a generation of graduates come through a degree, and they go and they go into the the workforce, and they go into the world, and people discover how little they have learned. And at that point, I suspect that some of the existing quality assurance bodies, which, you know, now we tend to have regulators of education or or people who, you know, credential schools and universities. And at that point, I think they might realize that something's gone amiss here. I think teachers, you know, again, the obvious way to resist this is to to start with, have a frank conversation with your students about what they will learn if they rely on these technologies, which is which is not very much. You you know, people do need to think about, as Jane was saying, the purpose of education, and we need to try to make the classroom not just about getting through materials so people can sit and examine, and I know there are many educators who, you know, feel deeply that they are trying to do something that isn't just about, you know, teaching stuff so that when someone can get a kind of piece of paper at at the end. Weirdly, though, in some ways, one of the obvious techniques for ensuring that students need to learn how to think for themselves and assess are assessed on how well they can think for themselves actually would be invigilated exams where people can't use use AI, and also the kind of, you know, talking to students or or assessing students' oral presentations that, you know, quickly this is one of the experiences that staff are having now, which is the student hands in a real what appears to be a really good essay, and you talk to them about it, and they are completely clueless because they haven't written they haven't written the paper. So maybe we should move to, you know, talking more with our students and trying to make that accessible in a certain way. Interestingly, this is a place where I actually can see a role for AI, at least in transcribing that conversation, Because one of the traditional problems with the oral exam or the assessment of, you know, a classroom presentation is a student can't appeal the grade, you know, and they can't necessarily they can't even go back and see what they did wrong. Well, now we can have a transcript. And so if someone needs it second marked and or if someone wants to go back and see, you know, assess their performance and see why they got the grade they did, they have we have some tools that we didn't have previously. You could just record it. You don't even need to transcribe it. But we need to be doing more talking with our students. We need to be doing we need to be setting assessment tasks that they can't complete using using AI. But, also, we need to just be open about what will happen if they overrely up upon these on these technologies. There there is a kind of hard question about how here about how sort of paternalistic educators should be. I mean, you know, part of me thinks, look, if a student wants to kind of get through their degree, never actually reading, writing, or thinking, they're adults, that's on them. But I'm also conscious that sometimes, know, students themselves want to be better than that, and being a little bit more interventionist and setting some hurdles that require them to make a start on that project of working themselves will sometimes benefit

Speaker 1 people. So that is that is also something we need to be thinking about. So can I just talk just briefly about the the students' own relationship to this technology? Right? So there was a recent study done at Monash and Deakin by Michael Henderson and Monash and and Margaret Berman at Deakin and some other people, you know, talking about how students feel about Gen AI systems. And overwhelmingly, they don't trust them. They trust their teacher more in the great majority of contexts. We also, with some people at Deakin, helped run a survey of of students using GenAI summarizers, something that Rob and I are very worried about. You know, students won't read. We've talked a lot about writing, but, you know, they can very quickly generate a 400 word summary. Adobe Acrobat asks you if you'd like a summary. You get confronted with this wall of text and, oh, dear, I've only got, you know, an an hour to to read it. Why don't I just get the summary? Students who use that technology are aware that they're losing something when they do so. Right? So they don't trust the Gen AI systems. They prefer to have contact with their teachers. And when they do use a summarizing AI, they're aware that they're kinda cheating themselves out of the experience and the kinda pleasure, of of reading and the kind of skill sets that are built, through reading difficult material. Right? So it's not a hard sell is what is what I wanna say. If you say to students, look. We're we're, we're going to teach I'm I'm gonna discourage the use of AI. I want you to think for yourselves. They're gonna say, yes. That's what I think in the majority of cases, they're say, yes. That's why I'm here. That's why I chose philosophy. That's why I chose literature. That's why I chose whatever discipline it is. Right? Because I wanna think for myself. And that's already coming through in the evidence that we're slowly building up as we talk to students about their experience of AI. They don't trust it, and they don't want to use it. They only use it when they're pushed for time. Right? And that has a lot to do about the way that our society as a whole is structured. It's the way it's organized. There's a cost of living crisis in Australia. They have to work a great deal, right, to pay the rent, and then they've got to get through their their classes and and meet all the assessment hurdles. But it's I don't think this is something that we have to, push on them. I think they already they want to be good readers and writers. They want to be able to talk to ideas. That's why they're engaged in the humanities. And we just have to find ways in the classroom of making sure that they can't, or that it's more difficult for them, right, to, at the last minute, in a in a panic, turn to AI. And so I I don't actually think we need to be, you know, paternalistic. Right? We just have to kind of enable them to do what they want, which is to become well, the majority of them want. Right? It's to become good readers and writers and thinkers.

Speaker 2 So outside of your own work, which of course I'll link in the show notes, where can you point listeners who wanna understand the ideas that inform your arguments? As in, who are you guys reading and listening to to inform the work that you're doing?

Speaker 1 This is a tough, question, Rob and I. Rob's been working for a long time on on machines' capacity to take responsibility, right, and and human oversight of machines, initially in a military context. Is that right, Rob? This was about automated weapon systems and whether or not we can hold a human being accountable for what they do. Right? So, you know, this this work on gen AI and responsibility is kinda natural kinda development or continuation of of what Rob's been working on for for a long time. And I know that he he has some a philosophic a philosophical background in in Wittgenstein and Raymond Gater, right, that's gonna inform that work. For for me, I'm a I'm a card carrying Hegelian. So the work that I read, I'm sort of in two places and trying to bring them together. So I'm reading the higher education literature because I want to know how what students are saying and what teachers are saying about their experiences with AI. And it's in a lot of cases, a know thy enemy kind of practice because there are a lot of people promoting AI or simply saying, okay. We're gonna roll these out, but responsibly, but not really thinking about just how constitutively irresponsible these systems are. Again, that's Robin, my intervention. But then I read the workers philosophers who are working in the post Segalian critical social theory tradition, which is it's something one can pick up, I guess. So I I read a lot of Robert Brandom and and Rahul Yaghi. Rahul Yaghi has a book that I'm rereading at the moment called Alienation. And I really think, that's what I wanna work on next. I really think that working with AI is a fundamentally alienated relationship. It alienates one from one's own thinking and one's own agency. Right? So that's you when I think about what what people might read to sort of get an understanding of where we're coming from, for me, at least, it's this kind of combination between that critical social theory tradition. So Axel Honnett, Radu Yaghi, also the work of Robert Brandom, connecting that up with what students and teachers are saying about, education. And and just, you know, that that recent paper by Michael Henderson and Margaret Berman and others, I'll send that to you as well to link because it's a really clear indicator of students, you know, themselves kind of, I don't trust this, which is what Rob and I have established philosophically, that they're fundamentally untrustworthy. They're not agents. They can't contribute to a genuine pedagogical context. And here's students intuitively feeling that. I'd much rather talk to my teacher about this.

Speaker 3 So I would definitely recommend the work of the British Australian philosopher Raymond Gaither, who has some very, I think, thoughtful remarks about the relationship between responsibility and embodiment and and our interpersonal relations in his book, good and evil, an absolute conception. That's pretty kind of heavy stuff, but for the sort of, you know, there's a there is a rich tradition of Wittgensteinian a thought influenced by Wittgenstein that I think is really important here. In some ways, I think some of the kind of older literature on, you know, critical pedagogy, you know, deinstitutionalizing schools, what the point of education is, I think people should, in a way, be reading some of the classics there because this is in part this is is about what we want education to be about, as we mentioned earlier. So there is a, you know, there's a long tradition of radical and critical scholarship where people have asked those questions about, you know, which agendas are schools and universities serving, what kind of lessons do they teach students about authority and about their own place in in the world? So, yes, in in some way, I I think some of the kind of older material on deinstitutionalizing schools, critical pedagogy is also really important today.

Speaker 2 Well, that's excellent. Thank you both so much for taking the time to talk to me today. No worries. Our pleasure. Well, my pleasure.

Speaker 4 Thank you again for listening to our podcast at Human Restoration Project. I hope this conversation leaves you inspired and ready to start making change. If you enjoyed listening, please consider leaving us a review on your favorite podcast player. Plus, find a whole host of free resources, writings, and other podcasts all for free on our website, humanrestorationproject.org. Thank you.