The Future of Architecture Search

This is where the genetic evolution analogy breaks down. You're saying actually with humans, we can do better than that. We can do bigger hops. And if you have like better principles of like architecture search, or you have like the AI generating really good ideas to do as well, because like whatever human is doing, you can get a training AI to automate it at some quality. But I'm not too excited about architecture search, even as a way of like scaling capabilities. Like, I don't know, it doesn't seem like that's going to be the first thing that causes AGI.