Speaker unverified: belief

Transcript context

…So do you think that can help solve hallucinations? Not solve it, but reduce it. Still, the LLM needs to know when to ask for a tool call. And second, it doesn’t mean the internet is always correct. You can do a web search for who won the World Cup in 1998, but it still needs to find the right website and get the right information. You can still go to the incorrect website and get incorrect information. I don’t think it will fully solve it, but it is improving. There was another cool paper earlier this year—I think it was December 31st, so not technically 2026, but close—on the recursive language model. That’s a cool idea to take this even a bit further. Nathan, you mentioned earlier it’s harder to do cool research in academia because of the compute budget. If I recall correctly, they did everything with GPT-5, so they didn’t even use local models. But the idea is, for a long-context task, instead of having the LLM solve all of it in one shot or in a chain, you break it down into sub-tasks. You have the LLM decide what is a good sub-task and then recursively call an LLM to solve that. And then adding tools—you know, each sub-task maybe goes to the web and gathers information, and then you pull it all together at the end. I think there’s going to be a lot of unlock using things like that where you don’t necessarily improve the LLM itself, you improve how the LLM is used and what it can use. One downside right now with tool use is you have to give the LLM permission to use tools. That will take some trust, especially if you want to unlock things like having an LLM answer emails for you, or just sort them. I don’t know if I would today give an LLM access to my emails, right? I mean, this is a huge risk. I think there’s one last point on the tool use thing. You hinted at this, and we’ve both come at this in our own ways: open versus closed models use tools in very different ways. With open models, people go to Hugging Face and download the model, and then the person’s going to be like, “What tool do I want?” Maybe X.ai is my preferred search provider, but someone else might care for a different search startup. When you release a model, it needs to be useful for multiple tools, which is really hard because you’re making a general reasoning engine, which is actually what gpt-oss-120b is good for. But on the closed models, you’re deeply integrating the specific tool into your experience. I think that open models will struggle to replicate some of the things that I like to do with closed models, where you can reference a mix of public and private information. Something that I keep trying every three to six months is Codex on the web, which is just prompting a model to make an update to some GitHub repository that I have. That set of secure cloud environment is just so nice for just sending it off to do this thing and then come back to me. This will probably help define some of the local open and closed niches. Because there was such a rush to get tool use working, the open models were on the back foot, which is kind of inevitable. There are so many resources in these frontier labs, but it will be fun when the open models solve this because it’s going to necessitate a more flexible model that might work with this recursive idea to be an orchestrator. Hopefully, necessity drives innovation there.…

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