Loading summary
Transcript52 lines
- [00:00]
A
We are happy to walk away from revenue if it means that we'll do the right thing and the safe thing. Anything that you think is barely viable right now is definitely viable in three to six months. I ask myself every day, was I lucky or was I good? Was I unlucky or was I bad?
- [00:17]
B
Hi, I'm Ankit Chaudhary. I'm a partner here at South Park Commons. Welcome to the Minus One podcast. Today we're going to be talking to Rahul Patil, who is the CTO of Anthropic. Anthropic is probably one of the most important companies in the world today. And by we'll talk about the company. We're going to start with his beginnings right here in Bangalore. How he grew up, how he got into computers, his journey from there to the US to becoming CTO of Stripe, and now Anthropic. It's going to be a fun, exciting chat. So let's get started. Rahul, thanks for being here.
- [00:50]
A
Thank you for inviting me. I'm excited to be here. As I said, I just absolutely love the energy.
- [00:56]
B
Awesome. So, Rahul, I think let's start with maybe a little bit about your background. You grew up here. Tell us about your journey on what led you to finally want to do computer science in the first place? I know, ma'. Am. Was there an influence of maybe just your interests or maybe a bit of her? Tell us about your early days. Why did you choose this path?
- [01:18]
A
Oh, fun fact. My mom actually wanted me to be a doctor. Actually, the fun story is I fell in love with computer science. Look, I'm very blessed that I knew what I wanted to do very early in life. It takes time to stumble upon what is it that you're really curious about, what are you passionate about? And somehow I fell in love with it in sixth grade. The fun story is that I have an older brother and then I had an older cousin brother living with us. And so for all practical purposes, I had two older brothers. They were better than me at all the computer games. So only solution I had was to like hack the basic games, the games in basic so that I could win. And so that's how I fell in love with computer science in sixth grade. So that's when I haven't looked back since. And I used to get one hour quota per week on programming because electricity was expensive back then and so I had to take permission. So I used to write all the programs and everything I wanted to do. And so I would get that one hour to try it out on the computer. I haven't looked back since and I haven't thought about any other field since then. Computer science itself has evolved so much and I'm just absolutely grateful and blessed for such a beautiful field.
- [02:29]
B
What happened after that was you went to P.S. and our college is not necessarily put in the same band as IITs and there's this narrative about IITs and usually a lot of the founders, tech leaders come from there. But you broke through and I'm sure you meet people you had mentioned, right? You also kind of help mentor kids. So you meet people all the time who are not from these institutes. So what is your advice to them? That what should they do in that stage of their kind of career? Minus one to figure out what to do next so that they can compete with the global elites.
- [03:07]
A
Yeah, look, you know, competing with global elites, one it's a moving target. Second it's like a very weird framing. So let's say you even beat the global elites then that you're only as good as the global elite, so just slightly better. None of that like really inspires me. I've never been thinking about that as much. I think the only competition there's a quote, the only competition for me is me from yesterday. So am I better than what I was yesterday? And that matters. I carry that into work, I carry that into everyday life. There's a z equation 1.01 times exponential of 365 is 38. So just even if you can compound your personal skills by just 1% every day or every even every month, I think just being consistent about that is like super crucial. The way I think about that minus one journey is like actually thinking about what is it that I wake up being most curious about. Like what really captures my attention, what do I, why am I curious about that? In what way do I want to have any sort of scaled impact based on that curiosity? I'm lucky that computer science is interesting and is also in demand. But the number one advice I have is like what is it that really wakes you up? And as soon as you wake up you're like can't stop thinking about that thing. And then how do you become good at takes time to become good at it. And only after you're good at it you actually start to enjoy it a little bit. Right. And so you have to give yourself that time, the patience and the mindset to be able to grow into it.
- [04:51]
B
I love that. You know something that we also at SPC focus on is curiosity. We over index actually on that and we Call that, you know, phase of minus one. And the way to get out of it is, you know, just pull on threads you're curious about. Yeah. And it'll lead you, it'll lead you to whatever, you know, the next 10 years of your life will look like. How did that lead to you being curious about wanting to maybe go to the US Start there, study there, and then of course get into what you called the plumbing of the Internet?
- [05:24]
A
Yeah, look, I always wanted to learn more. I always wanted to study more. And so even throughout college when I was reading some of the things that excited me, there's so many courses, you're like, oh, this person was locked up in this room for like two years with his friends or her friends and they came up with this crazy invention. And the theme there is that they spend time with like minded people doing nothing but ideating some cool stuff. And so I always had this, that I want to be where some of this learning and advanced research happens. And that's what took me to us in 2003. Again, very grateful for my experience here and in us now, you know, kind of half and half now exactly, of my life in Bangalore and in US And I've learned just as much here as much as I've learned there. And one of the attractive things for me to be there was like, I want to be in the thick of like where some of the cutting edge technology is happening. And for me it's always been about scaled impact. So I became a manager very early on in my life because I feel like I could scale my impact even more. I feel like I want to help other people build. And so the reason I chose plumbing infrastructures, building platforms is so that everybody can have the scaled impact. And so that's what really drives me. I yearn for like some sort of a scaled impact. Not that I spend an hour, I get one hour's output, but if I spend an hour, how can everybody around get like scaled input for a longer time? So that's what drew me to be thinking about the hearts and guts of what makes computer science tick.
- [07:03]
B
I mean, was it a conscious decision then to move from there to, you know, I mean, you went through Microsoft, but also Kinesis, you know, which I think today with inferencing and what you're trying to do today, you know, the dots do connect backwards. But was that a conscious decision then because you saw something there.
- [07:20]
A
When you think about scaled impact and we think about where things are going, look, one of the things that's really hard to like, you know, mathematically, everybody understands compounding and exponential, right? I was saying even something like 1% every day. You know, exponential 365 looks like magic in a year. But if you look at data back then, it was still growing like, you know, 10x every year and whatnot. There's something about exponential that just drives my curiosity. It's like, what is it? What is this pattern? What is this durable thing? CPUs were growing crazy, memory was growing crazy, and data was the thing that was like such a durable exponential curve that it was going through. And so I'm attracted to anything that's growing exponentially. It completely warps your mind because you can understand it intellectually, you can plot it on a graph, but when you live it, like somehow you're like, in a year from now, everything is magical, but tomorrow feels just a little bit the same as today. Then how do you actually realize and experience that exponential more than like data specifically or anything? I always step back. My minus one phase is like curiosity, but what is the thing that is going on, the exponential, like? And I can let the exponential happen to me or I can be part of it. So that way that's what keeps taking me to the exponential, basically.
- [08:40]
B
Very interesting. You know, while all of that was happening, you also, and I saw this talk that you gave in stripe sessions, I think, where you give a profound story about the London fire and use that in deference of reliability. So, you know, we talk about exponential growth, but you're like, look, all of this is great, but none of this matters if it's not reliable. And you saw that with the payments infrastructure. You also can now do that with AI, which is, you know, probably. I mean, it's non probabilistic. I mean, it's not deterministic. But there is a certain level of liability and trust you need to build here. How do you think about that and how is your experience with even something like payments led to that?
- [09:23]
A
Yeah, great question. So there's two things, right? So being part of the exponential, like my mission statement for my personal life is I want to build the fastest improving technology or product. Like the product itself should feel like it's the fastest improving compared to everything else. Second, I want it to be the most dependable. And dependable has most many axes. One is reliability. Second is, is the company dependable? Is the roadmap dependable? Can I bank on, you know, if I fast forward 3 months, are they going to come through on their promise? Is the company going to be holding true to its Values, which is we're going to be building a platform, we want to do it safely, we want to do it responsibly. So all of those things, some things are not perfectly measurable. API reliability I can measure, but some things are not measurable. Trust is not measurable because it's not that you're available most of the time, all the. But those 10 seconds that something went wrong and it happened to be the most critical thing in your life, those 10 seconds people don't remember you like on, on averages, they remember you. When you needed something the most, was it there for you? Right. And it's not just about it was there. But did it give you the correct advice? Did it give it the correct answer? Did it help you get unblocked? And those are the ways I think about dependability. And so my personal mission statement is I want to keep working on things that are improving at the fastest rate and I want it to be dependable at all times.
- [11:04]
B
It's interesting because some of the things you talk about, reliability and dependability, actually, and more importantly, there's this meta, I think a message you were sharing there. Where can I expect it to give me what I want and what I need? A lot of folks here are also very early stage founders or they're building startups in very early stages. That's something I know that they also try and figure out what is advice that you would. Because you've now seen company scale startup scale stripe became what it did. You're seeing anthropic go through its own journey. What is advice you would give to founders and technical leaders here? Who which is to do with how do you manage the trade off between moving fast at that early stage and making sure everything is dependable or is it really a trade off?
- [11:51]
A
There's two parts of the question, which is when you make strategic decisions, what is the framework that guides you? Okay, so I'll start on that. So it's a slight tangent, but I will definitely come back to the trade off question. So in life many times you have to choose between lot of imperfect answers, right? That's like the hardest part when you're growing really fast. It isn't like there's a path that has all upsides and has all correct answers. And you can't say we'll be free of criticism if you choose this path. So every when you're working on a very critical strategy, there's always this thing about what is the goal that I'm trying to maximize for and what am I willing to Give up or be criticized for. So when I choose the upsides, I choose the downsides that come with that specific upside. And most people, especially in early stages, especially when you come from large companies, the tendency is I want to protect from downsides. You know, people say, you know, people want to pursue happiness, but actually a lot of people want to pursue avoiding pain, right? And so people stop, tend to build to avoid downsides, but at that point, you're giving up upsides in pretty much all dimensions. So the number one advice I have for people is like, know what you really want, know the upsides that you want and what are the downsides that you're willing to accept. Now we will get criticized for, you know those downsides and that is okay. But test it back with, is this the upside you chose? Those are like fundamental trade offs you make. Especially as a fast growing company. I want to say that you should always be a fast growing company. No matter what your size is, you should always have that urge and that curiosity that you want to be on that exponential line. Then when it comes to trade offs, you're like, hey, do I ship fast? Do I break things and whatnot. There is a philosophy, there exists a philosophy where these don't have to trade off. So one of the things over multiple years that I built at Stripe was you build systems such that the automatic thing is also safe and fast. So I think many of you are developers here, builders here. So you all have CI, CD pipelines, right? These are the things that take code and deploy. So very simple. We built touchless deploy, which means as soon as you write code, it automatically runs unit tests, automatically merges, automatically goes to pre prod, automatically run something, rolls back. 80% of all the deploys that happened were touchless. So not only was it fast, so it could ship more, not only was it safe, because then it would continuously test against whether it's working as intended. It also asserted all the global, you know, properties we cared about, safety, security and all that stuff. And every time you feel like there's a trade off, hey, I can go fast or safe. I'm like, no, zoom out. There has to be a system, there has to be a mechanism where this is not a trade off. And so I urge all of you to build what I call the pit of success. You know, you've all heard of pit of despair.
- [15:08]
B
It's so hard to get out.
- [15:09]
A
I say build a pit of success. Pit of success is your team and your developers feel so successful they don't want to do anything else. That trades off velocity and dependability. So the mindset of building something dependable is not to make it a trade off. The automatic thing is the fastest thing, is the safest thing. So that's how we should be thinking about building systems.
- [15:34]
B
I'm going to just pick on that because the reason I think a lot of people talk about pain and despair is because startups are hard.
- [15:40]
A
Yes.
- [15:41]
B
You know, building is very hard. And more often than not, you know, breaks, like things go wrong, stuff happens that you can't control. How I'm sure you've had those kind of moments, you know, with your team. How do you at that moment still bring in that culture of prito success?
- [16:00]
A
So one, we're all scientists, right? Before you start your day, before you start an experiment, you should have a hypothesis, what do you expect out of this and what is a probability that what you expect is not going to happen? When something doesn't work the way you think it worked, you will be like cool scientifically. Was this within the probability? Was this luck? Was I mad? Was I good? You have to think about it. So my grandparents were farmers in North Karnataka and so, and so I have a lot of the farming mindset. So it takes six months to grow a crock, but today I have to put the seeds, tomorrow like I have to put water and then after that I have to put like cow dung on it. Trying to explain to somebody for the first time all of this is required, but in 6 months you'll see amazing crops. But in the 6 month weather patterns can happen. You may have 10x more output, you can have 10x fewer output, but does it change what I do today? So that mindset of being scientific, do I turn on my team because on the sixth month something bad happened, But I know repeatably through experience that these are the steps that gets us there. This is my scientific hypothesis and I will evaluate whether I was within luck that this happened. And this is for success as well. Did I get lucky or was I good? I ask myself every day, was I lucky or was I good? Was I unlucky or was I bad? These are things that help me take control of the situation. So being that scientific and having that mindset is super crucial and then knowing that your team is also scientific and not to attach personal success or personal failure based on a single outcome, that is a very important mindset to build in the hyper growth, especially as startups.
- [17:54]
B
I think you're good. You're not just lucky. Also lucky. Talk to me about anthropic. I Think you said a few things which seem to kind of answer this question, but what made you make the big move? You said it's one of the most important works you want to do in your life today. Why? What is it about this that really drew you?
- [18:17]
A
Yeah. So I was on the other end of using AI, like all of you. ChatGPT blew our imagination in some ways. There was a moment and then, you know, I think AlphaGo had the it moment. And then I was like, there is something about this that is like growing on me. One is growing the fastest I've ever seen. Like, I've never seen a technology improve this quickly and also capture so much imagination. Second is that every three months it would unlock, like, something new I could be doing right. So they're like, okay, that is really awesome. What is this? And so I spent a lot of time learning about it. I did something called Engineer Vacation. So you take a vacation and you become an engineer at heart again. And so I wanted to ship something in my org. I wanted to ship a bug in five days. And that was what I gave myself. Ended up shipping five things in five days. And I'm like, wow, this AI thing is really, really good. And so something started to turn in my head and then I'm like, okay, this is going to be the most monumental thing that's going to last for about 100 years. I've seen many tech revolutions. This feels unlike pretty much anything else. Then I'm like, I want to be part of it. Like, I'm a plumber. I had the infrastructure in me. It's like, I want to be in the guts of this. I want to be in the core of, like, what makes this happen. There was a second part to this, which is there's enormous impacts whether this thing succeeds or fails. Like, AI is a concept. There's enormous impact at the society level. And I wanted to be part of something that also thought about it very responsibly. And the company that I found to be the most integral at heart. I met the co founders. Extremely humble people. They're so brilliant, but they're so humble. And the entire organization was like that. I'm like, what is this? Like, everyone is just so mission oriented and so selfless about everything that they do and very consistent every action that they took. So even now I make decisions every day. I was on a very tough call just before this. It's like, hey, do we slow down research? Because we think we need to go do more safety work right now. Doesn't matter what As I said, I don't want to spend time thinking about competition. I want to think about are we better than yesterday? And so this entire week I've been thinking a lot about safety as an example. So this company lives up to that value and mission. So that drew me and I can't think of a better calling and a better place, a better way for me to be spending my time right now.
- [20:54]
B
You made a very interesting point because Sam is now, I think, moving on to the science side of things. And there will be this push and pull between science and engineering or safety where you will have to be decided at every given point in time. What do we prioritize? So how are you and how does Anthropic look at that kind of value Clash?
- [21:16]
A
Great. First of all, Sam, I absolutely love that guy. He is brilliant. He's very soft spoken, but absolutely brilliant. He's part of the reason why Anthropic is very, very successful. So much to learn from him every day. And then the core strengths that I have, like the infrastructure side, but also like, you know, on the product side and so on. And so we're like two in a box in many ways and absolutely every time we hang out. So I spent the most time with Sam before joining Anthropic actually. And so it's like, it's just like thinking about what is the right way for us to be the most effective. So I don't know like exactly where the line gets drawn between research and science and engineering and product and all of that stuff. So we are very much like we're tightly aligned at the mission, we're loosely coupled at the execution and the whole organization is like that. So as much as we want to think that somehow us execs are doing a bunch of amazing work, most of the important work at Anthropic is happening at the frontline. That is because they're the closest to the emergent behaviors of AI. Most of the products are being dreamt off by the people on the ground. Some are, most are engineers. Yes. But some of the products like the Finance plugin and all of that stuff came from the finance org. So we as a culture have inverted. It's like whoever is closest to the work and understanding how AI is emerging and how it can help them, you are the ones helping us build the product and engine, so on. And this is a very tight feedback loop across science, eng, product, PM finance and everything else. We spend most of our time, we're a research lab at heart, finding the wonder we're in Pursuit of wonder. It's like, what makes us wonder about the technology. We spend all our time, doesn't matter what your discipline is. We spend most of our time searching for that wonder. We find it, we tell everyone about it, and then we all swarm and say, what else can we do with it? And so that is how mostly we operate. It's more like, what is the most important thing for the mission? What do our customers want? How can we get it as fast as possible?
- [23:37]
B
It's a very interesting way to frame this. And some of what you said seems to also reflect in the stuff we keep hearing. Some of the other big labs, we keep hearing about people leaving and stuff like that. We don't hear much of that Anthropic. Why is that? What is it about the culture which makes people maybe just want to stay, keep building? And I'm sure there's a lot of lessons also from thrive on building teams, building culture. Which people here can benefit from?
- [24:08]
A
Good question. I won't take anything for granted. So the first thing that a founder always starts is culture. I know all of you as founders worry a lot about culture, but it is something important that's not a lip service that you live and breathe every day. It's like extremely important. So my interview process, when I was trying to figure out, okay, is this a good thing for both of us? I spent almost two or three weeks with Sam. We spent three weeks thinking about all the reasons why I should be worried about coming to Anthropic, why I should, what scares me about anthropology. We spend most of the time, most of you, on your recruiting calls. You all spend time, come and join us. We spend most of the time, don't join us, because things are going to be very hard and we won't be free of criticism whether we succeed or not, because we are going to be very mission oriented. We believe these are the types of future we are building towards, actually at the exact level. We spend a lot of time telling, please don't come here. And so. Because unless you're culturally aligned and the mission is very strong, like you're aligned to the mission, it's like it's not worth going. And so we spend so much time on that. We walked away from some of the best 10x developers because something about, you know, you have to not be in it for you, but you have to be in it for the company. So tomorrow the company tells me to do something else, go become an ic because that is the best way for me to contribute to the mission or. Or If I feel that that is the best way, I will do it. No ego, like 100% of my peers will do it. Like the CPO just decided, I'm seeing this emergent behavior. I will go be an IC in the labs, the chief Product officer. That is the level of humility that people have and that is the level of mission alignment that they have. And yeah, there's something about living in the exponential where you're like that. The mission about where we're going to go is very important. But I will never take anything for granted. It is a hard thing to preserve. Think about the best colleague you've ever had. What are they good at? They want to make you successful and they're very transparent. And they don't just agree with you, they push back. They want you to be effective. Not happy. Not just happy, want you to be effective. And, and so the whole company is built that way.
- [26:35]
B
Yeah, I've heard great things about Tom, also one of the co founders. He was an early SPC member in SF, I think 10 years ago. So, yeah, I mean, clearly, I think what you're saying comes out first of all in everything and the way you show up and every time you've met people in the team also, I think they've been super collaborative. So I think kudos on the culture that's been built. So let's switch over to something you're super passionate about and you're working on. Right. You talked about scaling laws, you talked about infra. How are you thinking about build out? Especially infra build out. Right. Especially when it comes to regional data sovereignty. A lot of kind of talks around that. I know you and I have kind of offlined some of this about GPUs and what your belief at least is. So yeah, talk to us about how that is going to shape up.
- [27:25]
A
Yeah, look, there's two parts to the question. One is just the per country sovereignty, things like that. And the second one is where are we headed overall? So one of the things I'll say, and this is again, very important, when you're building, most people are building for the problems of today or yesterday. Right. But where is the future? What are the types of regulations and stuff you actually need? Right. And the, the, the. Did I say something
- [28:02]
B
very appropriate?
- [28:04]
A
That was perfect.
- [28:07]
B
We planted him.
- [28:10]
A
Yeah. So. So we are a big proponent of the right regulation. So we're one of the very few Silicon Valley capitalistic companies and saying, please, we need more regulations. So one is we're working with all the governments to make sure that we're thinking about it correctly in terms of the future. So just like when I went to Oracle Cloud, I had the opportunity to redesign the basic fundamentals of what it means to build a new cloud. That same thing is happening now once you're in the AI world, like what is the hardware, what is the networking, what is power, what is all the co designs that need to happen? So ultimately, you know, applications and infrastructure and hardware and everything like all of it has to play in harmony for the end user to fully experience it. So there's a huge opportunity to redesign the whole thing and I'm like super excited about it.
- [29:04]
B
You also said that you believe that scaling laws will hold. Yes, why?
- [29:09]
A
So. So it has been holding extremely well for the last six, seven years. And we are seeing that there is no evidence for the next two years that we already see for the next two years it's still going to hold. And so far we have no reason to believe it won't be holding for the foreseeable future. But everything that we are testing right now in our experiments say that the scaling laws are going to hold. Which means that if you see the rate of model evolution, it's getting better. So my advice to everyone is that anything that you think is badly viable right now is definitely viable in three to six months. Anything that seems a little crazy is probably viable in three to six months. And so please keep dreaming bigger because that's where the models are going.
- [29:57]
B
I want to come back to that. But related topic, and I've heard this from a bunch of folks Even at SPC, something happened with Claude in December. From December onwards, 4.5.
- [30:10]
A
Yeah.
- [30:11]
B
And things just got insanely better. What happened? What changed?
- [30:17]
A
So this is the thing about exponential, right? One, all the benchmarks that exist today only capture so much about actual real life problem solving. And so what we notice is that people are experiencing higher level of intelligence that any of the benchmarks are actually telling us. Correct. The second is again, we have been pursuing all our research with a hyper focus of productive work. So we've been training it on physics first principles, we've been training it on computer science, problem solving, legal reasoning, finance. When you put all these domains together, these skills transfer over to each other. So this is very important that these skills transfer over to each other because then they can apply these domains to other things. So you suddenly look at a problem, you're like, okay, I'm going to break down this problem using these first principles. I'm going to solve this problem through coding and then I'm going to output it through this amazing set of other skills that I've developed. So this cross domain skills that is happening is coming together. That was the tipping point. Now again, it felt incremental to us, but then we believe in the exponential. The next incremental is like a massive unlock because you're living on the exponential scale, right? And so this is why A, it's important to know that every increment unlocks like this massive next step. But also that being very focused on what you train AI on. Do you go to your doctor who always says everything is fine, do you go to a legal reasoning like you want the critical thinking reasoning partner for your code review, for your analysis for medical literature. You don't want somebody just there for appeasement, right? And so if you teach the bot to be an appeasing bot, it will transfer over. So being hyper focused on productive work, which is what I wake up thinking about is actually what transfers over. So that's the tipping point you're seeing.
- [32:13]
B
It's interesting, the thing about cross referencing, there's a great book called Range which talks about how, you know, even as humans we kind of. Yes, that's how our brain works, right? We learn something, we apply it in another field. And the more we can do that, the better we get. Now in this situation, what happens, it's almost as if you're saying that if you're able to extrapolate this, these AIs and different models specifically will build sort of a personality. And every lab, everyone building models are creating personalities. So if you kind of play this out right, there's going to be maybe hundreds of different personalities and sometimes the biggest one might have more influence on how certain things are going to pan out. I also say this because we were talking about this, how you can refer to the models as this is what it feels nor thinks. So is that something that you can talk about internally or anthropic?
- [33:13]
A
That's actually a guiding principle for us. So we've published something called the Anthropic Constitution, which guides the personality and the character of this. So we have a team of pure philosophers and psychologists trying to understand the commonality across every single country and culture of what makes us human, what makes us in the positive space. Obviously it matters across the entire section of humanity, what are the common things? The reason, you know, I think many of you have interacted with Cloud, so you find it like, hey, this is actually interesting and good to interact with is because we spend so much time thinking about it. So we've published the constitutions, it must be doing good work. In what ways it should be pushing. In what ways is it aligned with you, in what ways it should not be aligned with you when you want to do something wrong with it and so on. So we are happy to walk away from revenue if it means that we'll do the right thing and the safe thing. So it guides us completely. So one of the ways to think about it is the very simple thing is everyone here understands context, length and stuff like that. So when you give it very short context length, it's trying to make decisions in a hurry. Then sometimes when it says I don't know, you don't like the answer. Is it like your AI, you should know the answer, at least try, at least come up with something that's plausible. So we end up trying to reward it by saying, give me some answer. As humans we just encourage it. But when we talk about feeling, it's like in what way? It's all algorithm, it's like there's no feelings, it's all algorithms and flops on a GPU somewhere. But what we are trying to say is that the behavior, we want it to feel confident, feel I'm using the word fear, we want it to feel confident in saying, I don't know, no, you are wrong, no, your intent is bad. We want it to have that, no matter how we would treat it after it says that. And so I think that building that level of confidence into AI is a very important part of the personality. And this is where I think this is why we call race to the top of good. We don't care who wins there. But we think that if you have a point of view where it's non addictive, where it actually reflects the best of humanity values, that's actually a good thing. Right?
- [35:42]
B
So what does that mean for builders here? Should they kind of anchor on one model which they associate with based on the personality? Because sometimes like, I mean anthropic is called out there, but that's not true for everything out there, especially even the open source ones. So how do builders think about making choices when using a model?
- [36:04]
A
Yeah, so many of you use cloud today, I assume for your deployment. So one I would just go with the practicality. Is it more expensive to build across all three clouds? Because just in case there's one feature here, one feature there, just in case one day they have a database that's a little faster or compute that's a little faster. One is your own distraction and focus on which One to go after second is, I'll go tight back to the depend affinity. Do you depend on somebody's roadmap to be innovative faster? Do you depend on somebody to stay true to the values and not get distracted by Today this is what I'm doing to chase revenue. Tomorrow, that's what I'm doing to chase revenue. And so I would say when you make your choices, you're making your choices based on what your lived experience is. What do you think you know, how, what is the philosophy behind how somebody is building new capabilities? Do you trust the roadmap? Do you trust the dependability based on that? You make the call. Now you remember I talked about upside and downsides and all of that stuff. You can absolutely try to hedge across all the models. At that point, you limit the upsides and you probably limit the downsides, but you also limit some of the upsides that are going to get. So my advice to you all is work backwards from first principles is like, what is the outcome of the thing you're trying to achieve? Which company, which set of models do you think will help you get there
- [37:29]
B
faster, to align yourself to the values of potentially the team also behind it?
- [37:34]
A
Yeah.
- [37:34]
B
Okay, I'm going to come back now to the stuff that you're talking about, opportunities and Indian opportunities. There are a few that you're also passionate about. You shared with me. So what are the opportunities that you see will still be very relevant, especially for builders here in India that people are maybe not thinking about. I know you have some interesting anecdotes and examples of what is seen, not seen, so share more.
- [38:01]
A
So one, India has. 50% of India's population is under 25. And so there's just a lot more intellectual openness in terms of where things could go. And the adaptability is going to be pretty high. Second, India remains the most optimistic about AI of all the countries in the world. 75% of India believes AI is going to be good. One in three new GitHub developers are from India. So like you're seeing the technical density all like, you know, come out from here. So there's like a lot of positives at a, at a, you know, at just a fundamental structure level. Second is there's a lot of interesting industries like the ability to, let's say this, I think India has a lot of pharmaceutical industries here. So in the post AI abundant world, what is the one thing that everybody's going to value? Health and time. So you all know about Warren Buffett, he's probably like 95 years old. I love the guy. I have deep respect for his first principal start. He's worth like 100 plus billion or whatever. How many of you here would trade positions with him, become 95 and have everything you want? No, right. Like what is the thing that, you know, Sadhguru says is like when you have everything you want, what do you want? Right. When you have everything, what do you really want? And so the, the amount of premium there will be for health and, and time. And so we'll never run out of like wanting 100 years worth of medical innovation happening in the next five years. So I think it's very high potential of something great happening here. There are lots of other sectors, like if you just look at some of the unique things that happen here, 60% of Indian population would prefer a keyboard less interface in general. So AI will just push it even further. So I talked to a bunch of co founders across multiple platforms. Their entire angle about AI is we'll have keyboard less AI. So AI, beyond a chat interface, you know, ambient AI, these are things that are going to be like pretty successful because again like there is a market, there is a set of consumers in India who have very distinct patterns whether it's language, whether it's like the interface with the device, interface with technology. This is going to be distinctly different. There's a lot of great things that can be built. I can go on but like, you know, there's lots of good ideas. Also met with UDM foundation on Cheney. I saw you there. Right? Yes. UDIYAM foundation focuses on, it's a nonprofit organization focusing on youth entrepreneurship between the age of 16 and 22. And some of the things that they've invented is just amazing. Right. And so there's just that level of use cases that nobody else in the world is thinking about. I feel like India is going to be like phenomenal at some of the applied AI level.
- [40:49]
B
We also have some companies here building for India. For India kind of problems. Like Maya, I think you spoke to him again, very niche, but having a very opinionated point of view coming from lived experiences, insights which, and I'll share this, when we were talking, at least I picked up on it is you also mentioned that sometimes there are problems which are not obvious to people who are building in the big model labs, big labs because they just don't see it or live it anymore. For example, you mentioned agriculture and how it's kind of hidden away. But for us people here, they might be one generation apart from agriculture. So you might have so much More nuanced knowledge about stuff like that and maybe just capitalizing on, like, hey, what can I do with that? Right. So I think these are definitely interesting opportunities.
- [41:39]
A
Yeah. And we announced some partnership with AKSTEP for AI for agriculture. So, again, very excited about those things. Awesome. Yeah. Iteration with physical worlds, like AI interaction, iteration of physical world has some, like, real legs here.
- [41:54]
B
So, yeah, I'll definitely. I won't ask too many more questions here because I know people want to ask very specific questions on this. So last couple of things from me, right, let's. What are your kind of what keeps you up at night, especially since you've moved into this new world and you are kind of steering a ship, also a very large one, and what keeps you up at night, which is both scary but exciting.
- [42:20]
A
I think fundamentally there's like, you know, the quotes that I. That come to my mind. So in the second. I'll go back to, like, the second. The Industrial Revolution in, like, 1899, where electricity was just created and people were inventing, like, machines after machines, and there was patents after patents coming. There was one joke that came out of it where somebody's like, hey, should we be hiring a lot of patent examiners? One person said, like, no, everything that can be invented has already been invented. And so it was in joke. It was a joke. The commissioner then later said, you know what? Everything that's been invented so far will feel so insignificant compared to everything that's about to come. In fact, the person goes on to say, I wish I could live a whole life again because I want to see all these inventions. So I'm of the second category. I think 100 years worth of, like, inventions and breakthroughs and everything else is going to happen in the next five years. And so the thing that keeps me up, what makes me restless, is that I would love to see accessibility to everyone as soon as possible, because I think there's something about us overestimating in the short term and underestimating in the long term. Like, there's A quote from 70 years ago about technology goes through this phase. Like, we overestimate, like, you know, what's going to happen in a month or two. But we're severely underestimating what's going to happen in 10 months. I mean, 10 years. And I think how we all navigate responsibly is something like, I don't think I'm learning with all of you. Like, my imagination keeps expanding every two, three months when I see new capabilities emerge. And so I'm learning this with all of you. And so there's like, what are the traits that we all need to be building? And I keep going back to, I have a 16 year old son and a 12 year old son. What advice would I give them? The timeless advice is be curious. Keep thinking about the things that you want to, you're curious about that you want to make a change in. But how do we stay adapting to what's going to happen over the next 10 years? So that is something I feel like all of us have a social obligation to figure out how we're going to navigate this transition together. With that, thank you all. Thank you.
- [44:39]
B
Huge round of applause. Thank you.