Nathan Lambert: belief

Transcript context

…So a few quick things. First of all, thank you for defining what you mean by tool use. I think that’s a great thing to do in general for the concepts we’re talking about, even things as sort of well-established as MOEs. You have to say that means mixture of experts, and you kind of have to build up an intuition for people about what that means, how it’s actually utilized, what are the different flavors. So what does it mean that there’s just such an explosion of open models? What’s your intuition? If you’re releasing an open model, you want people to use it, is the first and foremost thing. And then after that comes things like transparency and trust. I think when you look at China, the biggest reason is that they want people around the world to use these models, and I think a lot of people will not. If you look outside of the US, a lot of people will not pay for software, but they might have computing resources where you can put a model on it and run it. I think there can also be data that you don’t want to send to the cloud. So the number one thing is getting people to use models, use AI, or use your AI that might not be able to do it without having access to the model. I guess we should state explicitly, so we’ve been talking about these Chinese models and open weight models. Oftentimes, the way they’re run is locally. So it’s not like you’re sending your data to China or to whoever developed the model in Silicon Valley.…

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