Lenny's Podcast: Product | Career | Growth
00:57:09 · Lenny Rachitsky, Emmanuel Ameisen
[00:57:09] S0: I think that, that to me makes me think [00:57:12] S1: that there's like some fundamental structure of like information that like these models are picking up on that in some ways, like we're picking up on and that, you know, and maybe like a crazier sort like simulation level that kind of is like regularity in our world. And so that maybe makes me think that like, if you were to have the power to simulate something like our world, actually, like you would land in sort of a similar situation, like there's just kind of this like basin of where things end up and they're just the sort of like easiest way for things to arrange themselves. And so in that sense, maybe, yeah, I'm sort of, I could believe that we're in a simulation. I think I'm not sure I fully buy it. The other reason I don't fully buy it is that even though I think the structure like lends itself to it, there'd be a question of like, why? Why are we like, like who's running this and for what? And that one I don't know about. [00:58:08] S0: Well, something Elon said, it stuck with me about this, that if you were to run a simulation, it'd be a really interesting to be like an interesting time in history or something that you're really excited about. And it tracks that this is a very interesting time in history. [00:58:23] S1: Yes. Yeah. Okay, that's a fair point. That's a Data fair [00:58:26] S0: centers in space and robots and artificial intelligence.
00:58:26 · Lenny Rachitsky, Emmanuel Ameisen
[00:58:26] S0: centers in space and robots and artificial intelligence. [00:58:30] S1: Yeah. Okay. I see it now. I see it now. Yeah. Somebody like recently told me, you know, they used to think that may you live in interesting times was was like a blessing. And they're not so sure now they're like, you know, the boring times are like pretty good too. [00:58:45] S0: Oh, man. And then there's an Emmanuel out there watching your thinking process in your J space. [00:58:51] S1: Oh my god. Yeah, hopefully I'm making him proud. He's not like, oh, I gotta rerun the simulation. We messed up this one. Shut it down. Oh, man. [00:58:59] S0: Okay. One last question randomly out of nowhere. So Swix, who, runs AI engineering, latent space podcast, awesome guy. You were asked him what to ask you. You were on his podcast a year ago and something completely from a different direction. He wanted to get your take on JEV, which is this hot new model that recently came out that is very fast making decisions. OpenAI, funny enough, just launched their own version like a week later. Thoughts on JEV in the space of quick decision based models?
00:59:32 · Emmanuel Ameisen, Lenny Rachitsky
[00:59:32] S1: Yeah, right. So my take on JEV is this sort of, you know, these kind of models, you give them some context and you give them a few options to choose between, and they just are very quick at choosing an option for you. To me, it's impossible not to hear about this and think of sort of system one thinking versus system two or something, right? So like a lot of what I've studied with these models is how they sample one token with like the cloud models. And so, can think of like the decision models are as pretty similar. So like take an LLM and sample one token. They're different important ways, but like at a fundamental level, can be like, oh, you're not letting the model sample and like reason, it's just you're like gut feeling, is the answer A or is the answer B? To me, that seems like something that's like that general frame of like, sometimes you'll just want a binary decision or like one out of 10 or whatever. It's just like clearly, I don't know, like both clearly useful and something that I imagine is like probably going to be a part of like intelligence systems, right? Where you have, for some stuff you need to think about a long time, you know, it's like, what's our roadmap for the [01:00:43] S0: next year? Need to think about for a while.
01:00:43 · Lenny Rachitsky, Emmanuel Ameisen
[01:00:43] S0: next year? Need to think about for a while. [01:00:45] S1: And for some stuff where it's like, you know, is this person who wrote us a review happy or upset? You kind of, you're happy with the sort of gut judgment. So yeah, think this I think this kind of research is cool. And I think it's also kind of different from the, you know, it's like a bit of a novel take on some of the LMs we've seen. So yeah, I'm pretty excited to see where this kind of stuff goes. [01:01:08] S0: So what I'm hearing here is this kind of gonna be a new foundational component, most likely of AI and LMs. And the system one, system two stuff is basically Kahneman's kind of, I have that book up there actually behind me. System one, just to keep it simple, just like kind of your amygdala, limbic system making just like gut level quick decisions like snake, oh no. Yes. Where you don't have to think long about it. It's kind of to protect you from bad things really easily and quickly. And system two is where we really think about stuff and process. Yeah. Exactly. Sweet. Emmanuel, is there anything else that we haven't covered? Anything else you wanna say or mention? Anything you wanna double down on before we get to our very exciting lightning round?
01:01:08 · Lenny Rachitsky, Emmanuel Ameisen
[01:01:08] S0: So what I'm hearing here is this kind of gonna be a new foundational component, most likely of AI and LMs. And the system one, system two stuff is basically Kahneman's kind of, I have that book up there actually behind me. System one, just to keep it simple, just like kind of your amygdala, limbic system making just like gut level quick decisions like snake, oh no. Yes. Where you don't have to think long about it. It's kind of to protect you from bad things really easily and quickly. And system two is where we really think about stuff and process. Yeah. Exactly. Sweet. Emmanuel, is there anything else that we haven't covered? Anything else you wanna say or mention? Anything you wanna double down on before we get to our very exciting lightning round? [01:01:48] S1: I'm definitely curious for folks takes always on some of the stuff that, you know, still an explorer didn't make sense. There's a lot more interpret experiments than we have the time to share, you know, models have a bunch of weird, weird little things about them. Like maybe I'll do a lightning round of weird little things, but that's okay. [01:02:08] S0: Let's do it. Absolutely.
01:02:08 · Lenny Rachitsky, Emmanuel Ameisen
[01:02:08] S0: Let's do it. Absolutely. [01:02:09] S1: Okay. Weird little thing one, we found that, you know, models can like actually answer questions. Obviously they can do math or they can, you know, you can ask them like, what's the color of a banana and they'll actually like, you know, if you look inside, you'll see all these intermediate but they have a separate mechanism, which is kind of like a gut check and which is totally different. And which is just like, as they reason, they'll have like some part of them that will flag like, something's wrong. So if a model like somehow says, you know, four plus five equals 23 or something, There's a part of the model that just like reads that and doesn't do the math, but it's just like that can't be right. Because like four plus five definitely under 10. And so they learn these weird kind of like hunches. They feel like kind of like a hunch. Maybe another one is you can tell the model, hey, do this task and think about this other thing. And then you can use interrupt to see if like there's intermediate representations that map to the other thing. And that works. You sort of see that it does think about the other thing. This is something that's really mysterious to me because there's no kind of training reason why why the model would have to learn that. So it's sort like another one of these weird behaviors. And actually, maybe I'll keep it at two. Think those two are good enough.