Lisa Randall: uncertainty

Transcript context

…Yeah, there’s several directions here. One is on the theorem prover side, Wolfram Alpha where everything’s much more precise, and we have large language model type of stuff. One of the limitations of those is it seems to come up with convincing looking things, which we don’t know if it’s true or not, and that’s a big problem for physics. Large language models are more or less generalizations of stuff that we have. There’s still breakthroughs in AI waiting to happen, and maybe they are happening and maybe they’ll be good, maybe not, but that’s not quite the same. Maybe in some cases, it’s just pattern recognition that leads to important things, but sometimes it could be something more insightful than that that I can’t even put my finger on. It forces us to… We don’t really understand how smart we are. We don’t understand how we think about things all that well, actually. But one thing is true though, we are a lot more efficient right now than computers and coming up with things, we require a lot less energy to do that. If computers figure out how to do that, then it’s going to be at a totally different ball game. Here are clearly kinds of connections that we don’t know how we’re making, but we are making them. That’s going to be interesting. I say we’re in early stages, but this is changing very rapidly. Right now, I don’t think that it’s actually discovered new laws of physics, but could it in the future? Maybe it can. It will raise big questions about what is special about humans that we don’t quite appreciate. There could be things that are like that leap of insight that happens, truly novel ideas, that could potentially be very difficult to do.…

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