The Pop Culture of "AI"

Earlier this year I attended an all-day symposium on “Teaching and Learning with AI” hosted by my Faculty. There were various sessions and workshops, led mostly colleagues of mine, exploring both the affective and pragmatic concerns about the current “AI moment.” There were talks on the environmental costs of datacentres, the crisis in copyright and creative fields, and some attempts to widen the horizons of how we think about “AI” beyond the institutional and especially software-hosted force-feeding that we all know well by now; at the same time, there were examinations of supposedly helpful “AI tools” and discussions of how to write a coherent “AI statement” for your course syllabus. In all, it was a potpouri of possible ways of confronting, interrogating, and negotiating an AI-mediated world.

I was struck, mostly, by how inadequate the whole discourse seemed. While there was a wide range of expressions, some theoretical and critical, some pragmatic and interested in practice, at no point did the discussions – some 40-odd faculty, staff, and graduate students – attempt to engage with the colossal scale of what we are currently being asked to navigate: the claims of a wholesale change in the way society will work; the agendas of the “accelerationists” and the “doomers”; the unprecedented venture capital investment (and subsequent effects on the economy); the grift; and the cultural/cognitive impacts on a generation of learners. This latter ought at least to have been the main thrust of an event so named, but only one brief talk, by a librarian who helps run the student writing centre, even got close to this.

I was further struck by the fact that the Faculty who hosted this (my own Faculty of Communication, Art, and Technology) is largely made up of media scholars of one sort or another – be they theorists or designers – and yet there seemed to be an almost complete absence of what we might call a “critical” response. One might have thought that the assembled collection of people steeped in the traditions of McLuhan, Innis, Hall, Foucault, Gramsci, and the Frankfurt School might have felt inspired to respond in a more systematic way. But no, and perhaps this is the essence of this moment in time: a moment of paralysis and speechlessness in the face of violent shifts in the public sphere, AI being just one overwhelming provocation.

It is hard to write “the history of the present,” as my colleague Roberto Dosil liked to say. And perhaps it is not yet possible to mount a cohesive theoretical analysis, let alone challenge, to the onslaught of the “AI” agenda. Perhaps that is yet to come. In the meantime, then, what can be said that is meaningful, beyond expressions of anxiety and dread, or tactical moves to accommodate what appears uncontrollable?

Librarian Julia Long, of the Student Learning Commons, spoke of students’ confusion about the various competing demands posed in this moment – the demand to use “AI” tools; to not use them; to disclose them; to plagiarize; to not plagiarize; to prove one has not plagiarized; and so on – in a moment when these young people are in the crucial stages of figuring out who they are in the world, of what they think of themselves, of their abilities, their contributions, their ambitions and desires. “Identity formation” was the academese shorthand I used in responding to this, a term that probably doesn’t clarify much. It isn’t that “AI” is interfering with their sense of themselves, with their sense of personal identity. No, the sense here is much broader and more foundational. I think of “identity formation” in the context of what education is for; out of a tradition of curriculum theorists like Paolo Friere, the idea that education is not a matter of knowledge, of the acquisition of knowledge or even the production of knowledge, but rather the development of knowledgeable identities: smart people, competent people, maybe even wise people (we can dream, right?) – that this is what we’re really doing in education. And if so, then the aggressive push to centre “AI” in the middle of everything we’re doing is at best an added burden on those processes, and more seriously a destabilizing force.

I am interested – have always been, if I believe the scribbles I put down in notebooks as far back as the 1990s – in the mythology of technologies, by which I mean the stories we tell ourselves about them in order to make sense of them and their place – their role, their usefulness, their threats and benefits – in our lives and practices. In the 1990s, this was for me about the advent of the global internet and personal computing, personal media. Over the intervening decades, I and many others have considered how we think about the rise of the Web as central infrastructure of daily life, about the advent of planet-spanning social media networks, about the way these technologies and the stories we tell ourselves about them shape politics, the economy, and the way we think about our relationships.

The current “AI moment” began, as far as I can figure, in November–December 2022 with the public release of OpenAI’s ChatGPT. Reportedly,^[1] the people at OpenAI saw this as a relatively uninteresting release and were taken off guard by the gigantic public response to it. For OpenAI and, I think, for many of us who were paying attention even a little bit – there wasn’t anything radically new in late 2022. The ‘machine learning’ approach that underlies these AI models was already decades old; the ‘transformer’ architecture for deep learning dates to 2017 at the latest; and Nvidia had been making chips for deep learning at least as far back as 2009; Emily Bender, Timnit Gehru, and colleagues published the now-famous paper, “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” a year earlier, in 2021. My grad students in Publishing were playing with text generation using OpenAI’s GPT2 – with an early chatbot interface – in late 2021. So what happened at the end of 2022 to so utterly transform the public imagination around the possibility of “AI”?

I think the answer is that OpenAI was already engaged in a massive public relations campaign. Perhaps the reaction of ChatGPT was a unexpected boost to this; or maybe the story that ChatGPT’s blockbuster reception was a surprise to them isn’t entirely true; or maybe they saw it take off and then leaned into it in a hurry. Those who were paying attention will note that around that time (both before and after ChatGPT’s release) there was an awful lot of news being made around “AI Safety”: the idea that an artificial sentience could pose a danger to humanity on an unprecedented scale. There is no question that these stories helped to drive massive investment to companies like OpenAI and Nvidia. And of course, if these technologies are powerful enough to threaten humanity, you’d better be on the winning side, unless you wanna be left behind. And OpenAI had always – since its founding in 2014 – made “AI safety” a key piece of the brand.

We can come back to the threat of an artificial superintelligence later; I want to linger on “AI” as investment opportunity. I feel that this is actually the real story – that the rise of “AI” in the public imagination is a story about venture capital first and machine learning second, rather than the other way around.

Sam Altman, for starters, is more than a startup guy; he’s the king of the startup guys. His early career has everything to do with Paul Graham and Jessica Livingston’s Y Combinator – perhaps the most successful tech startup incubator/accelerator: to the extent that founder Paul Graham made Altman President of the company in 2014 when he stepped down. Altman founded OpenAI in 2015 with backing from people like Elon Musk, Peter Theil, and Reid Hoffman – and steered it from not-for-profit organization to “capped for-profit” (2019) to really-for-profit (2025). This was not a research-based startup; it was a finance-based startup. And as a financial entity (instrument?) it has succeeded wildly, the centre of the largest accumulation of investment capital in human history. Microsoft’s $1billion investment in 2019 was just the beginning: OpenAI was valued at $500billion in October 2025. But that’s just OpenAI; they were the early leader, but they are certainly not alone in this game. John Lanchester gives us a sense of the scale of this:

In the deep historical past of 2018, Apple became the first public company in the world to have a market capitalization of more than a trillion dollars. Today, each of the ten biggest companies in the world is worth more than $1 trillion. Only one of them, the Saudi oil monopoly, Aramco, has nothing to do with the future value of AI. The top company, Nvidia, is worth $4.45 trillion.

What happened in November 2022 was not the release of a groundbreaking new product; it was an improvement on what had gone before, to be sure, but the much more sensible interpretation is that in late 2022 a massive promotional campaign kicked in: one that would utterly restructure financial markets around the world.

Lanchester, in that same LRB article, goes on to explain that those ten biggest companies in the world are collectively worth $25.6trillion; that $15.1trillion of that has been accumulated since November 2022. Lanchester says, “Is it a bubble? Of course it’s a bubble. The salient questions are how we got here, and what happens next.”

Indeed, what happens next? You can see for yourself how a group of people have put a lot of money into promoting a particular answer to that question: the answer is, “use AI for everything!” – and you can see this by the nauseating prevalence of that damned pixie-dust icon in bloody everything. Every popular piece of software we use, from Google to Zoom to Word is asking you if you’d like AI assistance, as if it were “fries with that.” It’s inescapable; or at least, it’s work to get around it. There’s no way of easily turning off the damned AI summaries, and so on. It’s in the tools you use, whether you want it or not.

The best explanation of this I’ve heard is that this is again about finance first: the big investment opportunity currently is in data centre development. The idea is that our AI-powered future is going to require orders-of-magnitude more data centre capacity (covering the entire planet’s surface, according to Ilya Sutskever, although hard to tell if he’s being serious or not), and so – like the proliferation of ‘investment instruments’ designed and created in advance of the 2008 subprime mortgage crisis, new financial instruments have been created around data centres. You can invest in a limited-liability, limited term instrument that funds a datacentre (at astronomical cost): effectively, a “data centre futures market.” But to shore up those futures, it is in interests of the people behind these things to drive demand for the datacentres; the worst thing that could happen is that no one cares enough use all this capacity we are investing in. Note: the active verb is “investing in,” not “building.” Remember, this is about finance, not about technology. And, as darkly narrated by MIT investment guy Paul Kedrovksy,^[2] because AI data centres literally burn through silicon chips (because of the intense energy requirements in model training), the mean lifespan of a GPU is only 2 or 3 years; the result is that this is not an infrastructure play like the development of railways in the 19th century – where, after the investment bubble burst, we were at least left with railways. Rather, data centres are consumables.

The trouble – and this is what I really wanted my media studies colleagues to talk about – is that everyone is treating the “AI” moment as if it were a technological advance destabilizing an otherwise business-as-usual economic and political environment. But the shoe is on the other foot. This is an economic (financial, speculative) and political advance (about centralization of power) in an otherwise technological mediated society. The cart is pulling the horse. The tail is wagging the dog. The speculators are in charge of history. Do we need to even mention Trump at this point?

What we are looking at it a “pop culture of AI” – it doesn’t matter what it actually does or doesn’t do; what matters is what we think of it, the extent we think about it all the time, that we change our decision-making because of it. The hype machine has already succeeded, wildly, at creating a pop culture of AI, and at telling us exactly how it should feel. That’s an amazing accomplishment. It’s also incredibly dangerous, for a whole host of reasons – and the economic bubble is just one reason. What we desperately need is some alternative thinking, alternative storytelling, alternative imagining about this moment in time (to be fair to my colleagues at the symposium, this is the opening that Gillian Russell and Fred Lesage tried to offer). Are we prepared to just accept the pop culture that is sold to us by Sam Altman and his friends, and that a million journalists are happy to regurgitate as they pretend to stay current? Is a counterculture even possible anymore?