Dwarkesh Patel: belief

Transcript context

…I can’t speak for the emphasis of anyone else. I can only talk about how we think about it. The goal is not to teach the model every possible skill within RL, just as we don’t do that within pre-training. Within pre-training, we’re not trying to expose the model to every possible way that words could be put together. Rather, the model trains on a lot of things and then reaches generalization across pre-training. That was the transition from GPT-1 to GPT-2 that I saw up close. The model reaches a point. I had these moments where I was like, “Oh yeah, you just give the model a list of numbers — this is the cost of the house, this is the square feet of the house — and the model completes the pattern and does linear regression.” Not great, but it does it, and it’s never seen that exact thing before. So to the extent that we are building these RL environments, the goal is very similar to what was done five or ten years ago with pre-training. We’re trying to get a whole bunch of data, not because we want to cover a specific document or a specific skill, but because we want to generalize. I think the framework you’re laying down obviously makes sense. We’re making progress toward AGI. Nobody at this point disagrees we’re going to achieve AGI this century. The crux is you say we’re hitting the end of the exponential. Somebody else looks at this and says, “We’ve been making progress since 2012, and by 2035 we’ll have a human-like agent.” Obviously we’re seeing in these models the kinds of things that evolution did, or that learning within a human lifetime does. I want to understand what you’re seeing that makes you think it’s one year away and not ten years away. There are two claims you could make here, one stronger and one weaker. Starting with the weaker claim, when I first saw the scaling back in 2019, I wasn’t sure. This was a 50/50 thing. I thought I saw something. My claim was that this was much more likely than anyone thinks. Maybe there’s a 50% chance this happens. On the basic hypothesis of, as you put it, within ten years we’ll get to what I call a “country of geniuses in a data center”, I’m at 90% on that. It’s hard to go much higher than 90% because the world is so unpredictable. Maybe the irreducible uncertainty puts us at 95%, where you get to things like multiple companies having internal turmoil, Taiwan gets invaded, all the fabs get blown up by missiles.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.