Unsupervised Learning · 4 min instead of 59 min · Apr 9, 2026

In-context learning may beat custom RL for most companies

Chief Scientist, OpenAI

Jacob EffronHost · Partner, Redpoint
Growth investor in Abridge, Ramp and Physical Intelligence; ex-Flatiron Health.

About this segment

Asked whether companies should run their own reinforcement learning, Pachocki says in-context learning is far more data efficient and will improve. He expects general harnesses to meet users where they are.

From “OpenAI’s Chief Scientist on Continual Learning Hype, RL Beyond Code, & Future Alignment Directions”, Apr 9, 2026. Watch the full episode

“There is a much more data efficient way of learning that we know, which is like learning in context.”

Jakub Pachockiat 13:05 · Unsupervised Learning13:05

Takeaways

Still build evals and gather examples for your domain
Replicating current RL pipelines may not be the right path
General harnesses should work across domains for a long time

Watch 3:34 from 12:28

In this episode

37:36Why OpenAI hides chains of thought to keep them monitorable5:56
12:28In-context learning may beat custom RL for most companies3:34
24:25Continual learning is what scaling GPT models was always for1:45Up next

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

12:32
I feel like I talk to a lot of companies, and, like, one of the main questions everyone seems to be asking these days is, like, should we be doing, you know, our own reinforcement learning, like, you know, take an open source model and, like, we have some data on a task that people do. We have evals because we know our domain pretty well. Like, is this something that makes sense for us to do? Who are like, should we just wait for the models to continue to get better at some of these things?
12:54
What advice would you give for the many builders that listen to the podcast as they think through the extent to which they invest on the reinforcement learning side? Reinforcement learning can be a very data efficient way to really improve the model at some sort of task. There is a much more data efficient way of learning that we know, which is like learning in context. And this is maybe the most fundamental way that people teach these models.
13:18
You just prompt them with like examples with with with instructions for what you want. I expect that Learning is going to get much better over time And so I think it definitely really matters that the models can adapt to your context They can adapt adapt to kind of the the kind of tasks you care about so I think that will be very important I'm not sure if like You know replicating the kind of current URL pipelines is going to be like the right way to go about it But yeah, it's definitely a problem that that we're thinking about Yeah, there's almost like yeah You still have to do the work.
13:47
Like, you still should, you know, figure out what the evals are that matter, gather the data, the examples. But, like, it may just turn out in the future. You're far better off just feeding that into this context than trying to, like, do anything on, you know, your own model. Yeah. I think that's quite plausible. And I think that, like, you know, obviously people have seen the success of tools like Codex, which I know, you know, you've obviously been a key part of and wondered, like, you know, hey, do we need to build, like, our own kind of, you know, should we build our own harnesses or our own ways of using these things?
14:15
or, you know, for our own domains, whether it's, like, you know, legal or finance or healthcare? Or do we kind of just, like, take the harnesses that the large models do and kind of use them within, you know, with the context that we have? Any thoughts around, like, that? Like, the implementation of the harness shouldn't really be a limitation for a very long time. I think we'll be able to get, like, much more general harnesses that people can use for all sorts of other domains.
14:42
I mean, I think codex is pretty good actually if you try using it for things beyond coding. That's so interesting. Like a much more general harness being something that's almost like adaptive to or like just works across whatever the, you know, specific set of tools you have in your domain or specific set of things you want to expose to the model. Yeah, I mean, I think, and you know, I think it's also worth thinking about like, you know, why, like, you know, what is kind of the kind of ultimate interface that we want to interact to the model with. So the model gives some, the models give some UI for this, right? They can build their own UIs. They can kind of do things that, you know, people would find very time consuming. But yeah, I definitely think there is also just like a lot of space to kind of enable the models to access
15:25
like the current interfaces that we use for people, right? So I think like we want to have, you know, AIs on Slack, for example, or that are kind of plugged into our context and are able to learn from it and able to realize these existing things, right? So definitely there is some mate in the middle here, but definitely I believe long-term, by default, the AI should meet you where you are. And if not, that would be because it has new abilities.
16:00
It's not because it has limitations.