00:01:09 · Aditya Agarwal, Srinivas Narayanan
[00:01:09] S1: Thanks for thanks for coming to SPC, Srinivas. First of all, I definitely want to say what a time to be alive. Right? Yeah. It's a crazy place of innovation. Just a few quarters back, we're talking about, you know, pre training, SFT, reward modeling. And today, we are talking about AI civilizations, right? Given all the chatter around AGI and OpenAI has also been very vocal about responsible custodianship, Talk to us about... Maybe to start off, how far are we from AGI? And I'll just throw the... I'll lob this in the room right away. [00:01:41] S0: Yeah. We live in incredible times. I think the word AGI is, like, pretty ambiguous, so you can all have your own definitions for that. But, yeah, we have we have models that are incredibly smart. They are able to make decisions that are pretty pretty surprising and pretty exciting at the same time and somewhat concerning too, which we can talk about later. But, yeah, I mean, the... You know, but depending on your definition of AGI, these models are already good, I would say, at doing lots of tasks compared to many humans. And the the issue is not just intelligence. I think a lot of the issue is about do the models actually have the right context to do the job, do they learn on the job, etcetera. So there are some still interesting questions, but I think we have come very, very, very far in core intelligence.
00:01:41 · Srinivas Narayanan, Aditya Agarwal
[00:01:41] S0: Yeah. We live in incredible times. I think the word AGI is, like, pretty ambiguous, so you can all have your own definitions for that. But, yeah, we have we have models that are incredibly smart. They are able to make decisions that are pretty pretty surprising and pretty exciting at the same time and somewhat concerning too, which we can talk about later. But, yeah, I mean, the... You know, but depending on your definition of AGI, these models are already good, I would say, at doing lots of tasks compared to many humans. And the the issue is not just intelligence. I think a lot of the issue is about do the models actually have the right context to do the job, do they learn on the job, etcetera. So there are some still interesting questions, but I think we have come very, very, very far in core intelligence. [00:02:25] S1: 100%. You know, we are all about the minus one to zero journey. I'd love to love to hear about your early years and your early influences in life. Right? We, of course, were talking just about it, that the love for science and math. And, course, if you're at SPC, you have to go to IIT Madras and you have to do computer science. So I think that was given and preordained. But talk to us about your pre influences early in life and what were the choices that you took to... Both on the early academic choices and career choices that got you here?
00:02:25 · Aditya Agarwal, Srinivas Narayanan
[00:02:25] S1: 100%. You know, we are all about the minus one to zero journey. I'd love to love to hear about your early years and your early influences in life. Right? We, of course, were talking just about it, that the love for science and math. And, course, if you're at SPC, you have to go to IIT Madras and you have to do computer science. So I think that was given and preordained. But talk to us about your pre influences early in life and what were the choices that you took to... Both on the early academic choices and career choices that got you here? [00:03:00] S0: Yeah. I grew up in Chennai. I think I've always been interested in math and science. So I was I was very interested in physics when I was studying for IIT. The contrarian choice would have been to study physics. Mhmm. I probably would have done it if IIT Madras had a physics program. I didn't wanna leave Madras. So I said... But... I said I'll study computer science. And I was always interested in computer science, but I would say physics was probably a bigger love at that age. But I enjoyed computer science tremendously as I was studying it as well. I was interested in AI as well coming to think of it because I took all the electives that existed back then. Believe it or not, I studied neural networks in nineteen ninety four, ninety five. It was interesting. It was exciting, but, you know, that was the AI winters. Right? So it wasn't really practical. [00:03:53] S1: We were almost having a renaissance moment around that time, right, at least in some academic circles around neural networks.
00:03:59 · Srinivas Narayanan
[00:03:59] S0: Nineties, yeah, it was sort of the winter. I think... Yeah. I did write a chess program back in 1994. It was not very good. And it's embarrassing to say being bad at chess coming from Chennai, by the way. So... But... Yeah. I mean, I was... I think I was always interested in AI, but it wasn't a real thing. I think I studied computer science. I studied databases in grad school. I worked in IBM research for a few years. The Internet bug hit, so I kind of felt like I should do something more exciting. So I've done two startups. The second one led me to Facebook. And even at Facebook, when I started, you know, the AI wasn't a real thing. So I worked on, you know, products having a lot of impact. But part of working on photos, you know, we were doing face recognition, and that was the seed of AI where I was exposed to it. And that... You know, this was back in 2012, and we were trying to figure out how to improve our face recognition systems. And we realized that there was something big happening in computer vision around that time when ImageNet happened. Yeah. So I got exposed to that. And, you know, and for a few years, I sort of still worked on other things. But then in 2016, I came back and said, okay. This is what I wanna work on full time. So I will always been interested, but I think there might be the moment. I was always also very practical that way. I always wanted to work on things that are impactful, not just interesting. And I think that combination probably in AI happened around that time. And so since 2016, I've been entirely focused on AI.
00:05:30 · Aditya Agarwal
[00:05:30] S1: It almost feels like at Meta, you worked through four companies. Right? Photos. You have Oculus. You have applied research, of course, the deep learning work around vision, right, around the same time that the AlexNet paper came out. So I'm just wondering, what is the process to gravitate to interesting problems? Or maybe how do you... Yeah. How do... What... What's your nose on these things?
00:05:56 · Srinivas Narayanan
[00:05:56] S0: Yeah. I mean, at Meta, we went through... You know, when we started, there was really... Mobile wasn't really there, so most of it was web. Four technological shifts or, you know, web, mobile slash cloud. ARVR honestly was a distraction for me. So I did that for a few years, but my heart wasn't really in it. So I actually came back to work on AI. And so AI was a very big technological shift as well. So I think for me, I've always been gravitated towards being at the bleeding edge and something extremely innovative, extremely exciting, but also extremely impactful. And I think maybe I've always been open to having a beginner's mindset. I'm willing to go back to the drawing room. It's not like, you know, I was born with AI. I didn't study AI for, like, twenty years. But just being to go back to the basics and, you know, bet on yourself being able to pick up and learn a new technology, I've been willing to do that, and I think that has served me pretty well. Then a lot of people are maybe maybe afraid of making a change. You know? You think of yourself as, you know, I'm a data scientist or I am a, you know, mobile engineer. And, I think you have to break these preset mindsets. And when something exciting is happening, just learn. Right? And so that has served me quite well.