体育与武术

"I think we will. I’m definitely a worrier both about AI and non-AI things, but humans do tend to find a way. I think that’s what humans are built for—to have community and find a way to figure out problems. And that’s what has gotten us to this point. I think the AI opportunity and related technologies is really big. I think that there are big social and political problems to help everybody understand that. I think that’s what we’re staring at a lot of right now; the world is a scary place, and AI is a very uncertain thing. And it takes a lot of work that is not necessarily building things. It’s like telling people and understanding people, things that the people building AI are historically not motivated or wanting to do."

"And I think that’s the thing that’s probably more worrisome is human-machine amalgamations. This enables an individual human to have more impact on the world and that impact can be both positive and negative. Generally, humans have positive impacts on the world, at least societally, but it’s possible for individual humans to have such negative impacts. And AGI, at least as I think the labs define it, which is not a runaway sentient thing, but rather just something that can do a lot of tasks really efficiently amplifies the capabilities of someone causing extreme damage. But for the most part, I think it’ll be used for profit-seeking motives, which will increase the abundance and supply of things and therefore reduce suffering, right? That’s the goal."

"But if you think of big industries like pharmaceuticals, law, or finance, I do think they at some point will hire people from other frontier labs to build their in-house models on their proprietary data, which will be another unlock with pre-training that is currently not there. Because even if you wanted to, you can’t get that data—you can’t get access to clinical trials most of the time and these types of things. So I do think scaling in that sense might still be pretty much alive if you look at domain-specific applications, because right now we are just looking at general-purpose LLMs like ChatGPT, Anthropic, and so forth. They are just general purpose. They’re not even scratching the surface of what an LLM can do if it is really specifically trained and designed for a specific task."

"And the model also self-corrects, and that was, I think, the aha moment in the DeepSeek R1 paper. They called it the ‘aha moment’ because the model itself recognized it made a mistake and then said, “Ah, I did something wrong, let me try again.” I think that’s just so cool that this falls out of just giving it the correct answer and having it figure out how to do it—that it kind of does, in a sense, what a human would do. Although LLMs don’t think like humans, it’s a kind of interesting coincidence. And the nice side effect is it’s great for us humans to see these steps. It builds trust, and we can learn or double-check things."

"But I do think there’s certainly an aspect where the GPU trajectory was all planned. But on the other end, it’s also a lot of lucky coincidences or good intuition. Like the investment into, let’s say, biophysical simulations. I mean, I think it started with video games and then it just happened to be good at linear algebra because video games require a lot of linear algebra. And then you have the biophysical simulations. But still, I don’t think the master plan was AI. I think it just happened to be Alex Krizhevsky. So someone took these GPUs and said, “Hey, let’s try to train a neural network on that.” It happened to work really well and… …I think it only happened because you could purchase those GPUs."