Your dichotomy between real and fake is wrong

September 2024

Philosophy

A viral tweet sparked a predictable cycle of discourse about AI and academic integrity. Students are using ChatGPT to write essays. This is cheating. It deprives them of genuine learning. It degrades educational institutions. We need to respond.

The thing that's wrong with this analysis isn't the conclusion. It might even be the right conclusion. The thing that's wrong is the assumption underneath it: that what students were doing before ChatGPT was real, and what they're doing now is fake.

Universities have been optimising for performance of learning rather than learning itself for decades. You study to pass the exam. You write the essay to satisfy the rubric. The skills required to get a first-class degree have always included extensive knowledge of what your markers want to see, which is not the same thing as understanding the subject. High-performing students have always been able to discuss their grades' worthiness better than they can discuss the subject matter.

When ChatGPT writes a student's essay, it's doing the same thing the student was already doing — producing output that matches expected patterns — it's just doing it more efficiently and with less cognitive effort. The student reveals, with perfect honesty, that they were simulating engagement with the subject rather than experiencing it. The tool makes visible what was always true.

Consider a Paris New Year's Eve celebration. A couple takes selfies in front of the fireworks, curates them for Instagram, posts them with a carefully selected caption. Critics say this is inauthentic — they're experiencing the moment through a screen, performing it rather than living it. But this assumes there was an authentic version available. Were they going to stand there in spontaneous spiritual communion with the fireworks, free from self-consciousness and social performance? Almost certainly not. They were always going to perform the celebration. The phone just makes the performance legible.

The same logic applies to AI and universities. The criticism "you're offloading your thinking to a machine" only lands if you believe the thinking was happening in the first place. In most cases, you're offloading the simulation of thinking to a machine that simulates thinking. The simulation was always what the system was measuring.

This doesn't mean nothing has changed. The scale has changed. The gap between what people can produce and what they can understand has widened dramatically. The credentials system's already-tenuous relationship with competence has been further strained. These are real problems worth taking seriously.

But the outrage framing — AI is corrupting genuine learning — misidentifies what was genuine. It lets institutions that had already abandoned their educational mission blame a new technology for a failure that predates it by decades.

The more honest question is: what would it look like to actually learn something? And why has that question been so systematically avoided?