Kyle Corbitt

speaker

658 appearances 1 recordings 1 series first heard Oct 2025 last heard 16 Oct

Kyle Corbitt’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 1 in all, peaking in Oct 2025 with 1.

Appearances

newest first · ▶ plays the moment

Latent Space: The AI Engineer Podcast · Why RL Won — Kyle Corbitt, OpenPipe (acq. CoreWeave) · 16 Oct 2025 podcast

But it was sort of like one of those big if true things.

Like, if that is true, if it turns out that like just doing RL on your task is

Just like something everyone should be doing, and it's and it's just, you know, teaching these agents continually, teaching them through experience is just going to be a huge benefit.

Then like being the first people working on that would be a really, really like awesome position to be in.

So that's how we thought about it.

Is like, you know, less than 50% chance, but really big outcome if not, if so.

I think since that time, and I've been very transparent with this, like with our team, and like when I'm talking to other people, like

I don't think the chance that that is the right approach is 100% yet.

I think that we're still in the process, even after going through this and and, you know, doing that of like figuring out.

But the probabilities in my mind are going in the right direction.

Like now I think they're actually like

Today, I was actually just thinking about this with another conversation.

I think that the chances that like everyone should be or you know, everyone who's deploying an agent at scale should be doing RL with it, either as part of sort of like a, you know, like pre-deployment or even like continuously as it's deployed, that that's like the pattern that that's gonna get to.

I'd say there's like a fifty five, sixty percent chance that that's just like the better thing to do.

And that's informed by kind of like our experiments working with customers.

So anyway, not a hundred percent, but like going all the way back to your question, like no, it was not obvious.

I it w it was an informed bet.

You know, it's it's still a bet, but one that I'm I'm feeling pretty good about right now.

Probably push back on that a little bit.

I don't think the math is actually that complicated.

Showing 181–200 of 658 · page 10 of 33