Scaling reinforcement learning, tracing circuits, and the path to fully autonomous agents
At the exact same time we're seeing an actual, real-life, non-theoretical explosion in AI capabilities, I've become much less worried about the prospect of a silicon god converting the universe into paperclips. My p(doom), as the kids say, has dropped off a cliff. The idea behind this post is to lay out these underrated arguments in one convenient place, and document exactly why I changed my mind.
If This, Then That, Except for When...
On a recent Dwarkesh podcast, Sholto Douglas, an AI researcher working on reinforcement learning at Anthropic said: “There is this whole spectrum of crazy futures. But the one that I feel we’re almost guaranteed to get—this is a strong statement to make—is one where at the very least, you get a drop-in white-collar worker at some point in the next five years. I think it’s very likely in two, but it seems almost overdetermined in five. On the grand scheme of things, those are kind of irrelevant timeframes. It’s the same either way” 1.