Menu
Sign In Search Podcasts Charts People & Topics Add Podcast API Blog Pricing

John Schulman

๐Ÿ‘ค Speaker
528 total appearances

Appearances Over Time

Podcast Appearances

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

Whereas people doing more like English writing and composition tend more to be like US based.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

So,

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

Yeah, and I'd say there have been times when we needed to hire different experts for some of our campaigns.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

Some of them are very talented, and we even find that they're at least as good as us, the researchers, at doing these tasks, and they're much more careful than us.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

So I would say the people we have now are quite skilled and conscientious.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

Right, you don't exactly need that because, yeah, you can get quite a bit out of generalization.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

So if you... Like the base model has already been trained on tons of documentation, tons of code with shell scripts and so forth.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

So it's already seen all the FFmpeg man pages and lots of bash scripts and everything.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

And it's...

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

So even just giving the base model a good few-shot prompt, you can get it to answer queries like this.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

Just training a preference model like for helpfulness will, even if you don't train it on, probably even if you don't train it on any STEM, it'll somewhat generalize to STEM.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

And like, so not only do you not need like examples of how to use F of M tag, you might not even need anything with programming to get some reasonable behavior in the programming domain.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

I would definitely, yeah, I would expect things to move in that direction.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

It's unclear what's gonna be the best form factor, whether it's like something that's, it's like a Clippy that's on your computer and helping you with something, or if it's more like a helpful colleague in the cloud.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

So we'll see which kinds of form factors work the best.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

And I would expect people to try all of them out.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

Yeah, I would expect more like,

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

Yeah, I would expect something like, yeah, the mental model of like a helpful assistant or helpful colleague to become more real where you can share more of your everyday work or have it like, instead of just giving it one-off queries, you would have a whole project that you're doing and it knows about everything you've done on that project so far.

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

You can tell it, it can like even proactively make suggestions like,

Dwarkesh Podcast
John Schulman (OpenAI Cofounder) - Reasoning, RLHF, & Plan for 2027 AGI

maybe you can tell it, oh yeah, like remember to ask me about this and if I've made any progress on it.