The Light Between Us · an interview, illustrated

The Copernican Mind

Terence Tao on what the age of AI is teaching us about our own intelligence
Future Takes · Futurology, four-minute excerpt · transcript from the clip's auto-captions, lightly edited for reading · source: youtube.com/watch?v=DRDoABHToEo
The panel beside the words draws the idea as it unfolds. Scroll.
0:01
Terence Tao

I think one thing that this whole era of AI is teaching us is that our idea of what intelligence is, is not really accurate.

0:08
Tao

The history of AI has been: here's a task that only humans can do. Maybe it is read natural language, or win at chess, or solve a math problem. And one by one, someone finds some AI algorithm that also does that. Okay: we can now recognize faces, we can now understand speech. But you look at how it's done, and it doesn't feel like intelligence. It was some trick: you just cobbled together these neural networks and you run some algorithm. We were looking for some elusive, intelligent way of thinking, and we don't see it in the tools that actually solve our goals.

0:49
Tao

But maybe it's actually because intelligence is not what we think it is. Large language models in particular became very successful, and a lot of what they're doing is just predicting the next token: predicting the next word in a sentence. And that doesn't sound like something which is intelligent. If you ask someone to improvise a speech with no preparation, and at every moment they're just saying the next word that comes to their mind, a stream-of-consciousness dump: you don't think that this could actually work. But it does, for LLMs. And maybe that's actually a lot of what humans do as well.

1:24
Tao

This response I'm giving you right now is not rehearsed: I didn't think about it that deeply. You can get very far just by being in the moment, and reacting. There are some things that actually do require careful, deep thought. But you can do a lot of what seemingly seem like intelligent tasks almost by instinct, or by autopilot. That's one thing the age of AI is teaching us: that maybe we're not quite as smart as we think we are, sometimes.

1:55
Interviewer

So do you think that what we thought was intelligence, what we mistook for intelligence, was wrong? Or that maybe the thing it is doing is intelligence, and we're actually surprised that our own intelligence is maybe that formulaic, and that simple?

2:15
Tao

Yeah. And I think maybe we just embrace it. Sometimes we think we're clever, and: you came up with a genius idea, but it's actually just something that you remembered from some half-buried conversation several years ago. That's still intelligence. It's hard for us to step back and see what intelligence is as an outsider, because it is too tied emotionally to our sense of identity. But now that we have this other model of intelligence that we can study a little bit more objectively, I think we're learning a bit more about ourselves at the same time.

what the panel showsTwo figures facing each other: for the first time, something that isn't us reflects a workable image of how we might work.
2:57
Tao

It's like in astronomy: we had the Copernican revolution. For a long time we thought that the Earth was at the center of the universe. Because it didn't make sense that the Earth was moving, spinning, going around: why don't we feel it?

what the panel showsPtolemy's sky: everything, sun included, wheeling around a fixed Earth. It matched how it felt to stand still.
3:12
Tao

But we realized over time that the Earth is actually just one planet in the solar system. The solar system is just one system in the galaxy. The galaxy is one among billions. And that's fine. It makes the universe more amazing: there is just so much out there to explore. So fine, we're not at the center of the universe. But that's okay.

what the panel showsThe flip: the sun takes the center, the Earth joins the orbits, and the view pulls back until home is one dot among billions.
3:34
Tao

Somehow, cognitively, until very recently we were still at the cognitive center of the universe. We were in the middle, and then there were animals, and then ants, and whatever. But now we're sort of one of many planets of intelligence. And that's also a good…

what the panel showsThe same flip again, one register up: minds as a system of worlds: human, machine, animal, colony: none of them fixed at the center. The clip ends mid-sentence, which feels right for a revolution still underway.
the cognitive center