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

John Schulman

👤 Person
528 total appearances

Appearances Over Time

Podcast Appearances

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

I don't know if it's a phase transition, but there's some capabilities that work at multiple scales.

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

Yeah.

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

Yeah, it's not totally clear what we're gonna see once we get into that regime and how fast progress will be.

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

So that's still uncertain.

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

I would say, I would expect there to be, I would,

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

I wouldn't expect everything to be immediately solved by doing any training like this.

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

I would think there'll be other like miscellaneous deficits that the models have that cause them to get stuck or not make progress or make worse decisions than humans.

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

So I wouldn't say I expect that this one little thing will unlock all capabilities, but yeah, it's not clear, but it might like some improvement in the ability to do long horizon tasks might go quite far.

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

Yeah, maybe there's some,

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

There's some other experience that human experts bring to different tasks, like having some taste or dealing with ambiguity better.

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

So I could imagine that if we wanna do something like research, like those kind of considerations come into play.

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

Yeah, obviously there's...

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

they're gonna be just sort of mundane limitations around like affordance of the model, like whether it can use UIs and obviously the physical world or having access to things.

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

So I think there might be a lot of like mundane barriers that are probably not gonna last that long but would initially like slow down progress.

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

Yeah, that's an interesting question.

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

I mean, I would expect that models will be able to use websites that are designed for humans just by using vision, like when the vision capabilities get a bit better.

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

So there wouldn't be an immediate need to change them.

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

On the other hand, some websites that are going to benefit a lot from AIs being able to use them will probably want to design to be better UXs for AIs.

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

I'm not sure exactly what that would mean, but probably assuming that our models are still better in text mode than reading text out of images, you'd probably want to have a good text-based representation for the models.

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

And also just a good indication of what are all the things that can be interacted with.