Moral Education and Critical Thinking
On Technology, Systems, and Agency
This year, I taught my first large undergraduate lecture course since the advent of LLMs. I reworked the assessment structure not so much to discourage use of LLMs, but rather to make their use a matter of indifference to me. (That said, I am an anti-LLM absolutist. That said, it's more important to me to not be a cop than to prevent LLM usage.) Of course, I gave reasons at the beginning of the year as to why students should not want to use LLMs. I also taught a great deal of writing process, and enforced the generation of process documentation. And finally, in the last unit of the course, a history of computational media, I taught a week on the history of AI research and tied it to the present of LLMs. We read from Joseph Weizenbaum's crucial Computer Power and Human Reason. These were third- and fourth-year cinema studies program students, with a great deal of media-historical and critical-theoretical training.
At the end of the year, meanwhile, in our department's consultation with the leadership of the undergraduate student union, I heard all about how students used LLMs all the time anyway. I do not think the reason students decided to use LLMs to do their thinking for them is because they didn't understand that LLMs are bad. This is partly because the vibe, when we discussed ELIZA and LLMs and Computer Power and Human Reason, was decidedly anti-AI. But it is also because the course material was extremely well-suited to producing a critical understanding of generative AI, grounded in the history of technological media and the history of critical-theoretical thinking about technological media. If students were following the course material at all, they were in a position to understand the material, ideological, and affective valences of any media technology, including LLMs.
The particular impasse of teaching (or studying, or working, or living) with LLMs is not that we don't understand how and why LLMs are bad (damaging to our environment, our economies, our commons, our humanity). It is that we do not feel as though we can do anything about it. The sense of inevitability of this technology is a specific effect both of the marketing of these technologies by self-styled AI firms and of the frankly tacky enthusiasms of the class of employers who wish to automate away their employees. Our students live in a world where "AI" is experienced both as obviously bad and also definitely inevitable. I assume we have all seen any number of thinkpieces where people talk about the incentive structure of higher education and LLMs (e.g.). Let me offer a steel-man version of this argument, as I understand it.
Higher education on the liberal arts model (which governs almost the entirety of four-year North American higher ed) took as its purpose the formation of a broadly well-educated citizenry. To the extent that higher ed was (and still very much is) a matter of social and economic class, the idealized version was that the system produced people entering the elite leadership class who were "well-rounded." The promise of "general education" or core curricula was that all graduates, regardless of the particulars of their program of study, would have absorbed a fairly high baseline of historical, critical, aesthetic, and scientific thinking. In its highest-minded articulation, this was a kind of moral formation for the meritocratic and technocratic elite that run democracies. And, in the post-WWII years, as higher education was democratized, this moral formation was itself democratized, understood as necessary not only for the elites but salutary for the entire demos. For a democracy to work well, according to this thought, for the people to be able to govern themselves effectively, citizens ought to be well-educated.
Even if this was only a regulative ideal, and never fully realized, since the Reagan revolution (starting in California in the last 1960s), higher ed has been reformulated as developing human capital. What was important in higher ed was the economic mobility it promised. It's not that much of a stretch here, more a shift in emphasis than anything else. As the North American economy shifted from manufacturing to service work, and increasingly knowledge work, higher ed's promise of preparing students for work only increased in importance. In other words, in place of training citizens (who govern themselves) for democratic life, higher ed now trains workers (who do emphatically do not govern themselves). The moral education of the liberal arts has waned in favour of vocational training for knowledge work. This has gone hand-in-hand with the defunding of public higher ed by governments across the continent. (Here, I am thinking both of Maurizio Lazzarato's The Making of the Indebted Man and Wendy Brown's Undoing the Demos.)
To be sure, higher ed has always served both functions. Many of the most august and storied institutions in North American higher edâHarvard, my alma mater, Brown, my current employer, the University of Torontoâstarted as vocational schools for clergy. As ever, moral education and the development of human capital sit uneasily side by side in higher ed. Students have been made to embody this unease. Faculty (and I get the sense that it's not uniquely but most emphatically research faculty in the humanities) demand students learn something both for its own sake and for their own sake. But the meaning of for their own sake here is not because it will serve you well as a worker, but rather because it matters to you. Because it helps you understand the world you live in, because it helps you understand yourself or others, because it touches you in some way. It may also have workerly application! But the point is that there's a dimension of everybody's life that exceeds employment, including political, aesthetic, spiritual, and intimate life. Increasingly, that dimension is understood as beyond the proper remit of higher ed. To my ear, many earnest defenders of the liberal arts (often couched as a "defence of the humanities") reduce this dimension down to critical thinking. They then emphasize both the special relationship the liberal arts have to it, and its necessity for knowledge work.
(Or, when I'm feeling saucy, I lean on Immanuel Kant, Hannah Arendt, Stanley Cavell, or Lee Edelman. When I'm feeling saucy, I say that kind of knowledge I have to teach does nothing. It is a kind of aesthetic effect. It must, like art, be purposeless. Like aesthetic judgment, it must be arrived at in freedom. It cannot be compelled, like a mathematical proof, in which understanding necessarily entails agreement. Instead, all I can do or say is: this matters to me in my world, I invite you to see if it matters to you in yours. Or, in full sass, I say I have no knowledge to give anybody; theory is about better ways of failing to understand something.)
But I have recently started to understand that critical thinking isn't the thing. It's a thing, a very, very important thing! But, apparently, you can give students extremely robust critical tools to understand why generative AI is bad. You can offer them historical, theoretical, analytical, and aesthetic ways of understanding the problems with generative AI. Students can even articulate the problems with the discourse about how AI is inevitable. They also understand that higher ed is now mostly about job training for positions that appear now only to be the management of LLMs. And so: they get that LLMs are awful, they're evil, they suck the joy out of everything, destroy their learning, both pilfer and pollute the commons. And then: they use ChatGPT to come up with questions about the material in tutorial (discussion section, for US Americans).
What's missing is what I want to call the moral dimension of the education. The emphasis on critical thinking involves a faith that knowing that AI is evil means knowing how to resist it. But we are almost all of us in North America in 2026 living in a state of hypernormalization. Hypernormalization was coined by the Russian historian Alexei Yurchak to describe life under the late Soviet Union: the system is not working, radically so, but like Wile E. Coyote, even thought we're out past the edge of the cliff, we are very studiously not looking down. (Or, to update that reference, we are absolutely not looking up.) In Adam Curtis's words,
[I]n the 80s everyone from the top to the bottom of Soviet society knew that it wasnât working, knew that it was corrupt, knew that the bosses were looting the system, [knew] that the politicians had no alternative vision. And they knew that the bosses knew that they knew that. Everyone knew it was fake, but because no one had any alternative vision for a different kind of society, they just accepted this sense of total fakeness as normal... [A]lthough we are not in any way really like the Soviet Union, there is a similar feeling in our present day. Everyone in [the UK] and in America and throughout Europe knows that the system that they are living under isnât working as it is supposed to; that there is a lot of corruption at the top. But when ever the journalists point it out, everyone goes âWow thatâs terrible!â and then nothing happens and the system remains the same.
Everything is fucked, it all feels awful, this isn't sustainable, I hate this timeline. But it also feels impossible to rearrange things. The evil is inevitable. Or, perhaps more to the point: because it feels impossible to arrange things, we are all, continuously, studiously, not acting on our critical understanding and the absolutely rancid vibes. The moral dimension here, that must go alongside the critical part, is how to act on critical understanding. How are we to behave in light of the critical understanding?
Often, this moral dimension feels much more practical, actually, than the critical dimension. One of the ways I have come to be indifferent to use of LLMs in my large lecture is to insist on process. I teach students how to write, starting with a weekly practice of "process writing." But the moral practice here, the thing that's difficult and necessary, is to resist using things that we know are bad. And an awful lot of students are failing that moral test. And a lot of professionals, too.
The task I feel like I have in front of me, now, is to really embrace the fact that the education I most want to provide is one that goes beyond critical thinking to developing the moral fibre necessary to live in ways that accord with critical thought. I don't
Two addenda to conclude:
First, on commencement speeches. The kids really hate AI. I can vouch for that! They do! The vibes when I taught Weizenbaum were great. Weâstudents and TAs and professor allâwere all angry, in all the same ways, with all the same targets. And yet, the evidence of teaching in large classes shows that, no matter how much students may hate AI, a lot of them are still using it. I get the sense that some of the intensity of the anger from students is grounded in the cognitive dissonance of their hypernormalized lives.
Casey Muratori has a nice video taking apart Eric Schmidt's recent commencement address at the University of Arizona. Muratori points out that the socio-economic-technical system we live in, the one that has made LLMs feel inevitable, the one that just feels wretched, was built in no small part by Schmidt. But when he talks about the negative consequences of the modern internet that Google played a huge part in building, those just happen. His speech betrays no sense of agency, responsibility, or accountability. Bad things just happen. And it's us little people's job to absorb these negative consequences of Schmidt's actions.
And second, on fucking Jeffrey Epstein. One thing that we do in critical theory classrooms (humanistic and social scientific) is talk about systems instead of people. It's not a grand conspiracy of evil men, but the function of a particular set of social systems. For example, in the cinema studies classroom, we read Louis Althusser's "Ideology and Ideological State Apparatuses" essay to understand the ways in which cinema functions as a technology to produce conformity. But this is not a matter of filmmakers being conservative. Instead, it emerges from Hollywood's economic and social structures.
But Jesus Christ, the Epstein files pretty much show that it really has been a grand conspiracy of evil men. (I haven't checked, but I reckon that Schmidt was in a few proverbial smoky backrooms.) I think our "critical thinking" classrooms might need to acknowledge this more. In particular, understanding that the contradictions we are being made to inhabit, and the bad feelings that come from a critical awareness of these contradictions, are not only "systemic." That is, not inevitable, not the result of the implacable operation of social and institutional machinery that cannot be moved or changed. Rather, a lot of what's awful about the present really is the result of individual agents acting terribly. Making the villain a system is demotivating; if what's bad isn't the result of somebody's agency, then what agency could change it? But naming villains, that can restore a sense of moral agency. And that sense of moral accountability for the students is what's necessary both to resolve some of the contradictions of hypernormalization (you can, indeed, decide to live differently and then live differently) and to develop a real practical sense that the sociotechnical systems that organize our lives aren't as inevitable as Eric Schmidt might want us to imagine.