I laughed when I first noticed a pile of theory trading cards on a friend’s kitchen table in Lahore. They appeared ridiculous. A brief paragraph about power and discourse was printed on the back of glossy little rectangles that looked like they would depict a Pokémon or a cricket player, but Michel Foucault’s bald, expressionless face was staring out. My friend, a PhD candidate who had twice given up on her studies, claimed that she trusted the cards more than ChatGPT. I assumed she was kidding. She wasn’t.
On the surface, the comparison seems strange. One is a tangible item that was created by a small press and edited by individuals who signed it. The other is a chatbot that can write a 2,000-word essay on biopolitics in less than a minute after being trained on what could be the whole internet. You would assume there was no competition. However, the more I probed, the more I began to understand why she felt the way she did.
It is difficult to sum up Foucault’s ideas. He writes in a sideways manner. He doesn’t actually define power; instead, he illustrates how it functions through clinics, prisons, and the subtle pressure of being observed. If you ask ChatGPT what Foucault thought, you’ll get something tidy and self-assured, which frequently starts with the sentence “Foucault argued that power is everywhere,” which is both technically correct and the kind of thing a first-year undergraduate would write before realizing they hadn’t said anything at all. In contrast, someone who obviously read the books edited the cards. In perhaps sixty words, the Foucault card on the table described the Panopticon, discussed discipline and punishment, and explained why being watched is only one aspect of surveillance; another is acting as though you might be.
According to a recent University of Pennsylvania study, almost 80% of users who bothered to use ChatGPT followed its incorrect responses. The term “cognitive surrender,” which is a clinical term for an emotion that most of us have experienced, was chosen by the researchers. When you ask the machine, it seems confident, so you proceed. Last winter, I observed that a friend had stopped pausing while using ChatGPT to edit an essay. She pasted the response after the cursor blinked. There was no turning back the pages of a book, no moment of uncertainty.
According to Ahmed and Mahmood’s Critical Discourse Analysis paper, ChatGPT softened the harsher edges of communism and subtly leaned toward capitalist framings when asked about contentious political ideas. That makes sense. The chatbot is not an impartial oracle. A courteous voice smoothed over the average of what it was fed. Of all people, Foucault would have laughed at this. His entire project focused on how institutions with interests produce knowledge. Though it speaks in the first person, ChatGPT is precisely the kind of organization he cautioned against.
Naturally, the trading cards have a bias of their own. Someone decided which ideas to condense into a paragraph, which quotes to print, and which thinkers to include. However, the bias is apparent. The editor’s name is visible. If you don’t agree with the card, you can return it to the shelf. Because the chatbot is designed to sound correct, it lacks the humility found in the format.
Observing all of this makes it difficult to ignore the possibility that we might be losing something minor but genuine—the slight reluctance to accept an answer. A pause is imposed by the cards. You read, flip the card over, and reflect. Despite its fluency, ChatGPT seldom provides you with that opportunity. It almost doesn’t matter if Foucault is best learned through code or cardboard. Whether we still want to learn on our own is the question.
