Nathan Lambert

Like Perplexity, Google, Meta care about this. I think OpenAI and Anthropic are purely laser focused on agents and AGI. And if I build AGI, I can make tons of money, right? Or I can pay for everything, right? And this is just predicated back on the export control thing, right? If you think AGI is five, 10 years away or less, right? These labs think it's two, three years away.

Like Perplexity, Google, Meta care about this. I think OpenAI and Anthropic are purely laser focused on agents and AGI. And if I build AGI, I can make tons of money, right? Or I can pay for everything, right? And this is just predicated back on the export control thing, right? If you think AGI is five, 10 years away or less, right? These labs think it's two, three years away.

Like Perplexity, Google, Meta care about this. I think OpenAI and Anthropic are purely laser focused on agents and AGI. And if I build AGI, I can make tons of money, right? Or I can pay for everything, right? And this is just predicated back on the export control thing, right? If you think AGI is five, 10 years away or less, right? These labs think it's two, three years away.

Obviously, your actions are – if you assume they're rational actors, which they are mostly, what you do in a two-year AGI versus five-year versus 10-year is very, very, very different, right?

Obviously, your actions are – if you assume they're rational actors, which they are mostly, what you do in a two-year AGI versus five-year versus 10-year is very, very, very different, right?

Obviously, your actions are – if you assume they're rational actors, which they are mostly, what you do in a two-year AGI versus five-year versus 10-year is very, very, very different, right?

I think OpenAI's statement, I don't know if you've seen the five levels, right? Where it's chat is level one, reasoning is level two, and then agents is level three. And I think there's a couple more levels, but it's important to note, right? We were in chat for a couple of years, right? We just theoretically got to reasoning. We'll be here for a year or two, right? And then agents.

I think OpenAI's statement, I don't know if you've seen the five levels, right? Where it's chat is level one, reasoning is level two, and then agents is level three. And I think there's a couple more levels, but it's important to note, right? We were in chat for a couple of years, right? We just theoretically got to reasoning. We'll be here for a year or two, right? And then agents.

I think OpenAI's statement, I don't know if you've seen the five levels, right? Where it's chat is level one, reasoning is level two, and then agents is level three. And I think there's a couple more levels, but it's important to note, right? We were in chat for a couple of years, right? We just theoretically got to reasoning. We'll be here for a year or two, right? And then agents.

But at the same time, people can try and approximate capabilities of the next level. But the agents are doing things autonomously, doing things for minutes at a time, hours at a time, et cetera, right? Reasoning is doing things for... tens of seconds at a time, right? And then coming back with an output that I still need to verify and use and try to check out, right?

But at the same time, people can try and approximate capabilities of the next level. But the agents are doing things autonomously, doing things for minutes at a time, hours at a time, et cetera, right? Reasoning is doing things for... tens of seconds at a time, right? And then coming back with an output that I still need to verify and use and try to check out, right?

But at the same time, people can try and approximate capabilities of the next level. But the agents are doing things autonomously, doing things for minutes at a time, hours at a time, et cetera, right? Reasoning is doing things for... tens of seconds at a time, right? And then coming back with an output that I still need to verify and use and try to check out, right?

And the biggest problem is, of course, like it's the same thing with manufacturing, right? Like there's the whole Six Sigma thing, right? Like, you know, how many nines do you get? And then you compound the nines onto each other. And it's like, if you multiply, you know, by the number of steps that are Six Sigma, you get to, you know, a yield or something, right?

And the biggest problem is, of course, like it's the same thing with manufacturing, right? Like there's the whole Six Sigma thing, right? Like, you know, how many nines do you get? And then you compound the nines onto each other. And it's like, if you multiply, you know, by the number of steps that are Six Sigma, you get to, you know, a yield or something, right?

And the biggest problem is, of course, like it's the same thing with manufacturing, right? Like there's the whole Six Sigma thing, right? Like, you know, how many nines do you get? And then you compound the nines onto each other. And it's like, if you multiply, you know, by the number of steps that are Six Sigma, you get to, you know, a yield or something, right?

So like in semiconductor manufacturing, tens of thousands of steps, 9999999 is not enough, right? Right? Because you multiply by that many times, you actually end up with like 60% yield, right? Yeah, or zero. Really low yield, yeah, or zero. And this is the same thing with agents, right? Like chaining tasks together each time.

So like in semiconductor manufacturing, tens of thousands of steps, 9999999 is not enough, right? Right? Because you multiply by that many times, you actually end up with like 60% yield, right? Yeah, or zero. Really low yield, yeah, or zero. And this is the same thing with agents, right? Like chaining tasks together each time.

So like in semiconductor manufacturing, tens of thousands of steps, 9999999 is not enough, right? Right? Because you multiply by that many times, you actually end up with like 60% yield, right? Yeah, or zero. Really low yield, yeah, or zero. And this is the same thing with agents, right? Like chaining tasks together each time.

LLMs, even the best LLMs in particularly pretty good benchmarks don't get 100%, right? They get a little bit below that because there's a lot of noise. And so... How do you get to enough nines, right? This is the same thing with self-driving. We can't have self-driving because without it being like super geo-fenced like Google's, right?

LLMs, even the best LLMs in particularly pretty good benchmarks don't get 100%, right? They get a little bit below that because there's a lot of noise. And so... How do you get to enough nines, right? This is the same thing with self-driving. We can't have self-driving because without it being like super geo-fenced like Google's, right?