Lex Fridman对话Sebastian Raschka和Nathan Lambert：2026年AI现状

• The following is a conversation all about the state-of-the-art in artificial intelligence, including some of the exciting technical breakthroughs and developments in AI that happened over the past year, and some of the interesting things we think might happen this upcoming year. At times, it does get super technical, but we do try to make sure that it remains accessible to folks outside the field without ever dumbing it down.

It is a great honor and pleasure to be able to do this kind of episode with two of my favorite people in the AI community, Sebastian Raschka and Nathan Lambert. They are both widely respected machine learning researchers and engineers who also happen to be great communicators, educators, writers, and X posters. Sebastian is the author of two books I highly recommend for beginners and experts alike. First is Build a Large Language Model from Scratch and Build a Reasoning Model from Scratch.

I truly believe in the machine learning world, the best way to learn and understand something is to build it yourself from scratch. Nathan is the post-training lead at the Allen Institute for AI, author of the definitive book on Reinforcement Learning from Human Feedback. Both of them have great X accounts, great Substacks. Sebastian has courses on YouTube, Nathan has a podcast. And everyone should absolutely follow all of those. those.

This is the Lex Fridman podcast. To support it, please check out our sponsors in the description, where you can also find links to contact me, ask questions, get feedback, and so on. And now, dear friends, here's Sebastian Raschka and Nathan Lambert. So I think one useful lens to look at all this through is the so-called DeepSeek moment. This happened about a year ago in January 2025, when the open-weight Chinese company DeepSeek released DeepSeek R1, that I think it's fair to say surprised everyone with near-state-of-the-art performance, with allegedly much less compute for much cheaper.

And from then to today, the AI competition has gotten insane, both on the research and product level. It's just been accelerating. discuss all of this today, and maybe let's start with some spicy questions if we can. Who's winning at the international level? Would you say it's the set of companies in China or the set of companies in the United States? And Sebastian, Nathan, it's good to see you guys. guys. So Sebastian, who do you think is winning? - Winning is a very broad term.

I would say you mentioned the DeepSeek moment, and I think DeepSeek is winning the hearts of the people who work on open-weight models because they share these as open models. Winning, I think, has multiple timescales to it. We have today, we have next year, we have in 10 years. One thing I know for sure is that I don't think nowadays, in 2026, that there will be any company that has access to technology that no other company has access to.

That is mainly because researchers are frequently changing jobs and labs. They rotate. I don't think there will be a clear winner in terms of technology access. However, I do think there will be, The differentiating factor will be budget and hardware constraints. I don't think the ideas will be proprietary, but rather the resources needed to implement them. I don't see currently a winner-take-all scenario. I can't see that.

At the moment. - Nathan, what do you think? - You see the labs put different energy into what they're trying to do, and I think to demarcate the point in time when we're recording this, the hype over Anthropic's Claude Opus 4.5 model has been absolutely insane, which is just... I mean, I've used it and built stuff in the last few weeks, and it's... it's almost gotten to the point where it feels like a bit of a meme in terms of the hype.

And it's kind of funny because this is very organic, and then if we go back a few months ago, we can see the release date and the notes, as Gemini 3 from Google got released, and it seemed like the marketing and just, like, wow factor of that release was super high. But then at the end of November, Claude Opus 4.5 was released and the hype has been growing, but Gemini 3 was before this. And it kind of feels like people don't really talk about it as much, even though when it came out, everybody was like, this is Gemini's moment to retake Google's structural advantages in AI.

And Gemini 3 is a fantastic model, and I still use it. It's just kind of differentiation is lower. And I agree with Sebastian; what you're saying with all these, the idea space is very fluid, but culturally Anthropic is known for betting very hard on code, which is the Claude Code thing, is working out for them right now. So I think that even if the ideas flow pretty freely, so much of this is bottlenecked by human effort and the culture of organizations, where Anthropic seems to at least be presenting as the least chaotic.

It's a bit of an advantage, if they can keep doing that for a while. But on the other side of things, there's a lot of ominous technology from China where there's way more labs than DeepSeek. So DeepSeek kicked off a movement within China, I say kind of similar to how ChatGPT kicked off a movement in the US where everything had a chatbot. There's now tons of tech companies in China that are releasing very strong frontier open-weight models, to the point where I would say that DeepSeek is kind of losing its crown as the preeminent open model maker in China, and the likes of Z.ai with their GLM models, Minimax's models, Kimi Moonshot, especially in the last few months, has shown more brightly.

The new DeepSeek models are still very strong, but that's kind of a... it could look back as a big narrative point where in 2025 DeepSeek came and it provided this platform for way more Chinese companies that are releasing these fantastic models to kind of have this new type of operation. So these models from these Chinese companies are open-weights, and depending on this trajectory of business models that these American companies are doing, they could be at risk.

But currently, a lot of people are paying for AI software in the US, and historically in China and other parts of the world, people don't pay a lot for software. - So some of these models like DeepSeek have the love of the people because they are open-weight. How long do you think the Chinese companies keep releasing open-weight models? - I would say for a few years. I think that, like in the US, there's not a clear business model for it.

I have been writing about open models for a while, and these Chinese companies have realized it. So I get inbound from some of them. And they're smart and realize the same constraints: a lot of top US tech companies and other IT companies won't pay for an API subscription to Chinese companies for security concerns. This has been a long-standing habit in tech, and the people at these companies then see open weight models as an ability to influence and take part of a huge growing AI expenditure market in the US.

And they're very realistic about this, and it's working for them. I think that the government will see that that is building a lot of influence internationally in terms of uptake of the technology, so there's going to be a lot of incentives to keep it going. But building these models and doing the research is very expensive, so at some point, I expect consolidation. But I don't expect that to be a story of 2026, where there will be more open model builders throughout 2026 than there were in 2025.

And a lot of the notable ones will be in China. - You were going to say something? - Yes. You mentioned DeepSeek losing its crown. I do think to some extent, yes, but we also have to consider though, they are still, I would say, slightly ahead. And the other ones—it's not that DeepSeek got worse, it's just that the other ones are using the ideas from DeepSeek. For example, you mentioned Kimi—same architecture, they're training it.

And then again, we have this leapfrogging where they might b

原文超出正文长度上限，此处截断——上游还有内容，完整版见上方「原文 ↗」。