Alexander Embiricos: evaluation

Transcript context

…Yeah, yeah, the modern way we communicate, swipe left to right and vertical feed. And then the Sora video, okay, so I see how this all connects now. I see. Yeah. To be clear, we're not building that, but it's a fun idea. I mean, in this example though, one of the things that it's doing is it's consuming external signals, right? I think the other really interesting thing is if we think about what is the most successful AI product to date, I would argue, it's funny actually not to confuse things at all, but the first time we used the brand Codex at OpenAI was actually the model powering GitHub Copilot. This is way back in the day, years ago. And so we decided to reuse that brand recently because it's just so good, Codex, code execution. But I think actually auto completion and IDEs is one of the most successful AI products today. And part of what's so magical about it is that when it can surface ideas for helping you really rapidly, when it's right, you're accelerated. When it's wrong, it's not that annoying. It can be annoying, but it's not that annoying. So you can create this mixed initiative system that's contextually responding to what you're attempting to do.So in my mind, this is a really interesting thing for us as OpenAI as we're building. So for instance, when I think about launching a browser, which we did with Atlas, in my mind, one of the really interesting things we can then do is we can then contextually surface ways that we can help you as you're going about your day. And so we break out of this, we're just looking at code or we're just in your terminal into this idea that, "Hey, a real teammate is dealing with a lot more than just code. They're dealing with a lot of things that are web content. So how can we help you with that?" Man, there's so much there. I love this. Okay, so auto complete on web with the browser. That's so interesting. Just like, "Here's all the things that we can help you with as you're browsing and going about your day." I want to talk about Atlas. I'll come back to that. Codex, code execution, did not know that. That's really clever. I get it now. Okay, and then this chatter, what is a chatter-driven development? No, this is a really good idea, but it reminds me, I had Dhanji on the podcast, CTO of Block, and they have this product called Goose, which is their own internal agent thing. And he talked about an engineer at Block just has Goose watch him with his screen and listens to every meeting and proactively does work that he should probably want to do. So ships to PR, sends an email, drafts a Slack message. So he's doing exactly what you're describing in kind of a very early way.…

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