Not All of It

2026-09-29 · essays

On Sunday I went to a talk by Dr. Alessandro Topa, the chair of the philosophy department at AUC, on the algorithmification of culture, which is his term for how more and more of the things people used to do with and for each other, like teaching, explaining, translating and writing, are now being handed to AI. As someone who works in tech and is also a student of philosophy, I find it a really hard topic to have one clear opinion on. He was arguing that the humanities have a duty to study this seriously, and I agreed with him, but what I kept thinking about on the way home had less to do with the humanities and more to do with my own days, which are mostly spent building things with AI.

I should say first that I like it, a lot. There's a real pleasure in describing something in a few sentences and watching it come out working, or close enough to working, in a fraction of the time it would have taken me a couple of years ago. Every few weeks something that felt impossible becomes routine, and learning how to get the most out of these models has honestly been one of the more exciting things I've done since I started working. I'm not writing this as someone who wants to go back.

But there's a feeling that comes with it that I don't really have a good word for. You finish something, it works, it's good, people are happy with it, and somewhere in the back of your head you know that you weren't fully there when it was made. You steered it, you checked it, you made the decisions that mattered (or the ones you think mattered), but the thing itself came out of somewhere else. The closest I can get is the feeling of looking at photos from a trip you don't remember very well.

And the uncomfortable part is that you can't really do anything about it. Not using AI isn't a serious option when everyone around you is using it and moving faster, and it would be irrational to pretend otherwise, because the productivity is real. In my own field it helps us produce more, faster and often better, and I genuinely believe that's good for the economy and for most of the people using it. So the feeling just sits there, and you keep working.

The usual answer when anyone brings this up is that AI is a tool, like the computer was a tool and the machines before it were tools, and that every generation panics about its new technology and then gets used to it. There is something true in that. Topa actually made a version of this point himself, talking about how every button we press is sitting on top of some practice we've forgotten. Where you now turn a knob to heat a room, someone once had to carry coal down to the cellar and keep checking whether the fire was still burning, and asking "is it still warm?" was something a whole household cared about. He was clear that he doesn't want the coal back, and neither do I.

What makes me less comfortable with the comparison is what's on the other side of the button this time. The things being automated now are much closer to thinking out loud, asking someone a question, or trying to explain an idea badly and then less badly, and the change is taking a couple of years where the heater took a couple of generations, so we're living through it while still trying to figure out what it's doing to us.

You can see it very clearly in a classroom. In most of my classes this semester, the majority of laptops have ChatGPT or Claude open, and when people get stuck, that's where they go first, before the person next to them and before the professor. I don't say this to judge anyone, partly because it's often the faster way to an answer, and partly because I worry I might be doing the same thing without noticing. But a classroom was never only about getting answers. Topa said something about this that I keep going back to, which is that putting a thought into words is part of how the thought becomes clear to you in the first place, so when you hand the half-formed version to a model, you skip part of the thinking along with the effort. When I look at those classrooms the answers are all there, and what I notice missing is people asking each other things.

What surprises me most is how little we talk about any of this in tech. We talk about AI constantly, but almost always about the models, the benchmarks, the new release that's better at coding than last month's. When the human side comes up, it's usually a joke, something like "well, we're all going to be unemployed in five years anyway," and everyone laughs and goes back to prompting. I've made that joke myself more than once. I think it's the only way the subject is allowed in the room, because taking it seriously would make it harder to open the laptop the next morning and do the job, and so there's a kind of unspoken rule that you can be excited about AI, or you can be sarcastic about it, but you can't really sit with it.

And when I do try to sit with it, the question I end up at is a simple one that I can't answer: what are we going to look like in thirty or fifty years? Jobs matter too, but what I keep thinking about is culturally and humanly, whether the people who grow up having never written a first draft alone or asked a stupid question in class will be the same kind of people we are, or whether we'll still be the same species, quote unquote. It could turn out fine. Maybe we'll lose some things and gain others, the way we always have, and people in 2076 will read posts like this one the way we now read Plato worrying that writing would ruin our memory. I honestly don't know, and I find it strange that so few of the people building this are even asking.

I don't have a conclusion, and I'm suspicious of anyone who says they do. I'm going to keep building with AI, because I think it's good and because I'd be lying if I said I didn't enjoy it, and I'm going to keep feeling that I'm not fully there for all of it. The least I can do is stop treating that feeling as something to joke away, and I was glad to spend my Sunday in a room where someone was treating it as a serious question.