{"url": "https://www.youtube.com/watch?v=nLMZothlRNM", "page_url": "https://www.youtube.com/watch?v=nLMZothlRNM", "title": "Mosaic Ventures in conversation with Y Combinator President, Sam Altman", "show_or_channel": "Mosaic Ventures", "date": "2017-03-28", "duration_s": 3945, "type": "interview", "speakers": ["Sam Altman"], "confidence": "medium", "notes": "Mosaic Ventures conversation", "people": ["sam-altman"], "id": "2017-03-28_mosaic-ventures-mosaic-ventures-in-conversation-with-y-c", "audio_duration_s": 3944.896, "worker": "jc-w15", "transcribe_s": 50.5, "model": "large-v3", "segments": [{"start": 21.46, "end": 61.54, "text": "Good evening. Good evening, London. Welcome. Thanks for joining us. Apologies for the security and the queuing, but unfortunately, you know, we have a rock star with us, so we had to put you through all of that, just to know that you were coming to something very special tonight. And there are, unfortunately, some people outside who weren't able to make it in, so that made you guys feel even better. But tonight, we're very, very lucky. My partner Simon and I are..."}, {"start": 61.54, "end": 89.94, "text": "I'm going to be in conversation with Sam Altman, who's a great friend of Mosaic and somebody who has really interesting insights into the world of startups and as president of Y Combinator. Sam's history, for those of you who don't know, Sam went to Stanford, dropped out, and did a startup, which he raised money for some top venture capitalists on the West Coast."}, {"start": 89.94, "end": 118.38, "text": "This proceeded to sell that company and join YC. YC is the preeminent accelerator in the world. And I think today it has backed over, I think it's 3,200 founders was in your annual letter, is the number. I think 1,470 companies. And the market cap of those companies is over $70 billion in terms of the ownership that YC has in those companies. It's pretty good for..."}, {"start": 118.48, "end": 154.46, "text": "A little startup that's been going for a dozen years. And I think Sam also mentioned that he thought that number could get over $100 billion by the end of this year if it has some tailwinds. So companies like Dropbox or Airbnb or Stripe, Docker, it's a great long list. So please welcome Sam Altman to the stage. So let's start off with one of your favorite questions, maybe just to warm things up."}, {"start": 154.46, "end": 179.55, "text": "We'll use this in the context to sort of talk about YC and what it's all about and what you guys stand for, but maybe just talk a little bit about founders and what you look for in great founders. I think you have a sort of strong opinion on what a great founder looks like and the types of people that you'd love to join YC. We have 50,000 people apply every year from around the world, and we get to pick the ones that are going to do this."}, {"start": 179.55, "end": 209.07, "text": "Very ambitious thing. Maybe fail. But if it works, build these massive successes. And so we spend a lot of our time trying to optimize for great founders. Because if we get to work with really good people, our job is very fun. And if we don't, it's incredibly painful. And so we thought a lot about this question. And one of the things that we've noticed a while ago is that determination was the most important characteristic of great founders. More important than intelligence, vision, anything else."}, {"start": 209.07, "end": 239.12, "text": "We want founders that are relentlessly determined to accomplish their vision. And evaluating that has been what we've had to tune our application process for more than anything else. We want founders who have a big vision. I think it is easier to build a hard company than an easy company. If you're building the 10,000th photo sharing application, it's very hard to get people to come join your mission. If you're trying to build a supersonic airplane or payments for the entire internet,"}, {"start": 239.12, "end": 268.91, "text": "People are very excited, and they want to be part of that. You can get customers, you can get investors, talk to the press, and most of all, you can recruit employees. And so we like people who have an ambitious vision. They have something that they want to see happen in the world, and they think a startup is the best way to do that. And this works for a lot of things. People are able to accomplish incredible things that people think are not part of the traditional startup mandate by being determined with a clear vision."}, {"start": 268.91, "end": 298.78, "text": "We like people who are good communicators. That is important all the way through for all those same reasons. We like people who are nice and that we like to be around. One of the things with a startup is that you end up spending a lot of time on it. And we spend a lot of time with the startups for fun. And so we like to work with people that we want to be around and that are energizing. And then finally, one of the things that I think is magic about Silicon Valley, and I haven't seen anywhere else in the world to the same degree, although I hope you all can create this same community here."}, {"start": 298.78, "end": 328.72, "text": "People believe in the future and aren't mocked for it. You can be incredibly ambitious. You can say, I'm going to go do this amazing thing. I'm going to start a website for college students that's eventually going to have a billion users. I'm going to organize all the world's information. I'm going to build a nuclear fusion company. I'm going to build a rocket company, an electric car company. I don't have anything today, but give me a bunch of money and I'm going to make this happen. If you say that in most places in the world,"}, {"start": 328.72, "end": 357.25, "text": "People either to your face or behind your back make fun of you. And in Silicon Valley, they've learned their lesson. And they at least stop and think, maybe I should believe her when I hear that. And this has been really powerful. And I think more than anything else, this is the secret to Silicon Valley outperformance. And it's also a ton of fun to be around. And so we look for people that have that spirit. This, I'm going to figure it out. I'm going to get it done."}, {"start": 357.25, "end": 385.57, "text": "And I believe this thing that most people don't, but it's really important, so I'm going to go do it attitude. And that, I think, is a distinguishing feature of the founders that we like to fund. Tom, you're also kind of on record for saying you really look for non-traditional founders, for unknown people. You don't like people being on a career track. Do you want to comment on that? Yeah. As startups have become more popular,"}, {"start": 385.57, "end": 414.78, "text": "It's now something that people do for a resume item. And Y Combinator in particular. You know, people sort of view their lives on this train and on the way to Stuyvesant and then Harvard and then Y Combinator. And those are the people that we want to run as far away from as we possibly can. The founders that we like are the founders that are doing this because they are kept up at night because they believe that what they're doing is going to be so great. And that is not a resume item."}, {"start": 414.78, "end": 445.26, "text": "You don't do that as a step to something else. You do this because it's the last thing you want to work on. If Y Combinator were the last thing I worked on, I'd be happy with that. If I did this for the next 50 years, I'd be happy with that. And you want founders like that. We have never had success funding founders that are doing a startup as a resume item. And we have built up all of these defenses to try to detect that. I think one of the distinguishing problems with university admissions processes"}, {"start": 445.26, "end": 473.25, "text": "They tell you, kind of like, here's what you look for, and there's this test, and you have to take it, and there are all these prep courses that help you take it, and you need this many extracurriculars, and all these other things. And that, any system like that, that you make hackable, will get hacked. But those aren't necessarily the people you want. And so we have intentionally designed our admissions system as best we can to be unhackable. Because we want people who are here"}, {"start": 473.25, "end": 501.26, "text": "Because this is the thing they want to do with their lives, not because they think it'll help them with their next job. For the companies that you fund, I think in the last cohort there was, I believe, something like 106 companies that went through last week. What's your expected hit rate, and how do you kind of embrace the fact that the majority don't work? About 50% roughly fail, eventually."}, {"start": 501.26, "end": 528.9, "text": "The other 50% have some degree of success. But one of the things that I think is maybe unique about Y Combinator is we really try to internalize the power law. And we understand that all of our returns, effectively, to a first approximation, come from the very best company in each batch. And so we have this rule, which is, because of that, the founder's interests come before anybody else."}, {"start": 528.9, "end": 558.08, "text": "More important than our relationship with VCs, more important than our own returns. And we are long-term greedy, but in the short term, we say, you know what, do the right thing. If it's not working out, give our stock back. Don't try to make a little bit of money around the edge. Because all the returns are from this best company. So to our mind, we don't think in terms of failure rates. We think in terms of the value of the most valuable company. And everything else, we just want to be nice and have a chance that those founders send us their friends and come back the next time. In fact, many of our most successful exits"}, {"start": 558.08, "end": 587.12, "text": "Have been from founders on their second or third time through Y Combinator. Our biggest exit of last year was this company called Cruise, a self-driving car company that GM bought. And that founder was on his second time through Y Combinator. And his first time through went really well, but it wasn't always going well. It had a few pivots and a few turns along the road. And I think had we not taken the approach of, you know what?"}, {"start": 587.18, "end": 615.86, "text": "We're just going to do the right thing for you. We wouldn't have gotten to invest in Cruise. And so I think most investors look at failure rates and also try to win even in the non-successful cases. And our view is make sure we have the privilege and the opportunity to invest in Airbnb and otherwise not worry about it. And if the company fails or doesn't do that well, we'll just help the founder. The founder can have whatever money they can get from it and we move on."}, {"start": 616.62, "end": 646.59, "text": "And from the founders' perspective, what we see increasingly is if they have YC on their LinkedIn profile, even if the company didn't work out, that's still a badge of honor and a rite of passage. And the failure is not a stigma, which it might have been 10 or 20 or 30 years ago, particularly outside Silicon Valley. Well, I think failure is still bad, and people should try to avoid it. And I think one of the odder things about Silicon Valley is like,"}, {"start": 646.59, "end": 675.73, "text": "That people brag about failure. Failure sucks. I have, you know, like as anybody else, I've done some things that have worked really well and plenty of things that haven't. And it's much more fun when things work. But I think Silicon Valley is very tolerant of failure. Failure is not a black mark. And this is another thing that I think we get right more so than most places around the world that really helps us. People are not afraid to try things because they know if they fail, it's no big deal."}, {"start": 675.73, "end": 705.07, "text": "And that has been a great asset to Silicon Valley. We did a survey once, and I asked a bunch of our founders that had failed, you know, are you still happy you did the startup? Because I worry about when I come to a room like this and try to convince a bunch of people to start startups, that I'm giving people bad life advice. And to my great reassurance, I think everybody except one person said they were happy, even though they failed."}, {"start": 705.07, "end": 734.59, "text": "They were happy with the experience, that they learned more than they would have any other use of that couple of years. They met great people, and they got a great job at the end of it, and it wasn't so bad. They obviously wish they had succeeded, but they picked themselves up, dusted themselves off, and moved on to the next thing. That said, I'm actually not here to convince any of you to start startups. You probably shouldn't. We try to have truth in advertising, and I think for most people, it's the wrong life choice. Probably for many people in this room, it is."}, {"start": 734.59, "end": 763.47, "text": "But starting a startup is really incredibly difficult. And there are many days where you really hate it. And we try to always be honest and upfront with that about people. Because the way that the media covers it, and especially covers YC, it looks like everyone we find is successful and everyone we find is happy. And that could not be any further from the truth. Well, at the same time, you've also written in your blog about the fact that if people are going to go work in a startup, they better offer it."}, {"start": 763.47, "end": 793.79, "text": "Doing their own startup they're going to work in another startup which is likely to fail and they'll end up working with people who worked in a failed startup who may not be the best people unless they can get a job in one of those amazing breakout companies that you've worked with. Help people here think about that kind of choice and also other choices they might have. I think there's a set of economic arguments and there's a set of career development arguments."}, {"start": 793.79, "end": 821.66, "text": "The economic arguments are well documented online, but in general, I don't think early employees ever get compensated with enough equity. And so in that sense, you may as well be a founder, but you can read the numbers online. The thing that I don't think gets discussed is how different the experience is as the founder of a company versus the first employee in terms of what you get to learn and the kinds of things you get to do. The thing that I loved about being a founder..."}, {"start": 821.66, "end": 849.73, "text": "And the thing that turned out to be really important to the rest of my career as an investor was that as the founder, you get to do every job. And as anybody else in the company, you end up focused on one vertical area. And as the founder, you get access to things only the founder gets to do. Raise money, talk to the press, set up other things, do really high stakes business development, sales, hire the executive team."}, {"start": 849.73, "end": 878.56, "text": "I didn't appreciate, I still never had a real job, so I didn't appreciate how different that was, because I had nothing to compare it to, until I talked to a lot of people who had had both experiences. And employee number two and founder, although they both worked incredibly hard and are both really important to the success of the company, if you start your own company, you get a career development opportunity that is extremely difficult to get any other way."}, {"start": 878.56, "end": 909.18, "text": "And when you think about the founder, quote unquote, is it typically a founding team? So what's the best composition? Is it one person, two people, three people? If you look at the data, what does it tell us? We went through our data on this. Our most successful companies had two or three founders. We are willing to fund solo founders. I think it's difficult, but not impossible."}, {"start": 909.18, "end": 954.35, "text": "We try to be open-minded. All right. The moment you start cutting off potential companies and having rules saying we don't fund solo founders, we don't fund five-person founders, that's when the one comes along that breaks the rule. And given this power law rule that I mentioned, and there's one company that we invest in every year that is responsible for half of our returns or two-thirds of our returns, that's like an unacceptable risk. So we will consider anything, and it is far worse."}, {"start": 954.35, "end": 982.88, "text": "To pick a bad co-founder than to be a solo founder. Infinitely worse. That almost always blows up. But if you have an opportunity to have a great co-founder, you should do that. Maybe talk a little bit about the newest initiative that you've launched, which is a way to extend YC's brand to particularly a global audience, including the folks here. You guys may be aware."}, {"start": 982.88, "end": 1009.49, "text": "YC's launched something called Startup School, which Sam's been very much behind and is really driving forward. Can you just talk about the objectives for Startup School? What's involved in terms of the commitment by YC teachers, advisors, and also what's expected back from the founders who participate in it? Well, let me talk about why we're doing it just for a minute, and then we'll talk about what it is."}, {"start": 1010.32, "end": 1040.69, "text": "I think one of the things that has gone really wrong in the world in the last few years, and you see a lot of evidence of this in terms of Brexit and Trump and the populism sweeping Europe, is that we no longer have the economic growth that we had for several decades. And I think this is really important for democracy to function. I think if people..."}, {"start": 1040.69, "end": 1070.06, "text": "People are much more sensitive to relative quality of life than absolute quality of life. They want their lives to get better every year. I want my life to get better every year. And in a world where the entire pie keeps growing, then everyone's still happy. Democracy functions because everyone's life gets better. In a world where the pie stays the same size and we have sub 2% real GDP growth, then we fight. Because for your life to get better, mine has to get worse. We're trying to reallocate the pie instead of grow the pie. And I think..."}, {"start": 1070.06, "end": 1103.1, "text": "This is one of the few great threats to the systems and the values that we care about. We're basically honorary English, you know, in the US that we all care about continuing. And I think it's really important to get economic growth back. And the thing that I think is most likely to deliver that is innovation and startups. And I think we want to spread the sort of message of this around the world."}, {"start": 1103.1, "end": 1132.75, "text": "Because we think a lot more startups is really good for the world continuing to function well. And I think that economic growth that benefits everyone, and we can talk about inequality later maybe, is really important. So we thought about what we can do, because right now we can only fund a few hundred startups a year. And I think it would be good for the world if they were like, we could help tens of thousands a year. And we thought about the ways to do that, and we decided the best thing we could do was try to rebuild"}, {"start": 1132.75, "end": 1161.73, "text": "in software as much of the Y Com in your experience as we could. So the dinner, the talks that we hold, the community, one-on-one advice from people. And we built this thing. It's called Startup School. We start next week. We're teaching it in Stanford, but it's being broadcast online. And a bunch of YC alums have signed up to advise the companies in the program. And we'll have communities online as well so people can interact with other founders. And we're going to try to replicate the program for free as much as we can to anyone around the world."}, {"start": 1161.73, "end": 1192.98, "text": "And I think it's just startupschool.org if you want to sign up. It should be, I think it'll be really effective. We got a great set of guest lecturers and advisors, and I think we've been able to replicate a significant portion of the YC experience. Is the video going to be canned so that you can go back and do an archive, or is it live-streamed and you have to kind of log on at that moment? It will be archived. You will be able to watch it later. But for people participating in the class,"}, {"start": 1192.98, "end": 1221.06, "text": "We kind of want you to watch it every week because there's sort of a sequence to it. And we're asking the companies to sort of do the same weekly check in a YC company would do. And we're going to have like a thing similar to demo day at the end for companies that complete the entire course where you'll be able to present online what you've built. How much direct interaction will the founders participating have with the esteemed group of advisors that are at YC?"}, {"start": 1221.06, "end": 1248.53, "text": "We're targeting once a week, which is the same that we target for YC companies. We've had a lot of trial and error, and I think the right answer is like eight or nine days. Every eight or nine days, you want to sit down and meet and report progress. And that will be in a large group or a small group of founders? So one sort of YC value is try things and error correct quickly, so we're going to try both. We'll have some people that do one-on-one meetings. The group meetings are really helpful."}, {"start": 1248.53, "end": 1280.03, "text": "Every week at Y Combinator, we have group office hours, which is like two YC partners and seven companies together, and the companies get to hear from each other, and that's really valuable. It may not translate as well online, but we're going to try that too. How many founders do you expect to participate in this first wave? Mid-tens of thousands, probably. If it goes well, though, we can scale it up a lot, I think, in future years. The limit on it is just the number of advisors that you have who can support the founders?"}, {"start": 1280.03, "end": 1309.71, "text": "Yeah, but whenever we can, we like to fight scale with scale. And if we graduate a bunch more good founders, we have a bunch more good advisors. You've got that cloning startup as well. Yeah, one of these days. And will there be any funding for any of these companies or some of the better ones maybe? How are you thinking through the other elements of the YC value proposition? In the first version, there won't be. There's just sort of like one moving part too many. In future years, I expect there will be."}, {"start": 1312.14, "end": 1342.48, "text": "And so you'll get a certificate at the end of this that you completed the course, and that will obviously... Yeah, I mean, you will get a certificate. I don't know if that will motivate anyone. What you will also get, though, is the opportunity to present to a wide audience of investors and press and potential customers and partners and everything else at the end. That's a big attraction. Got it. Is there e-sports spectating allowed as well? Funnily enough, you are not the first person to ask that. The answer is no."}, {"start": 1342.48, "end": 1374.35, "text": "You see these other accelerators do reality TV shows and stuff, and that makes us cringe so much. I can't imagine any good founder wanting to do that. In the same way that we tell startups that they should write code and talk to users and do almost nothing else, or build their product and talk to users, our version of that is that we should recruit startups, decide which ones to fund, and advise them, and almost nothing else. And all the esports,"}, {"start": 1374.35, "end": 1397.86, "text": "Reality TV show stuff just feels wrong. Sure. I was joking. Other people haven't been. One other initiative that you also announced recently was about this AI track that you've started. I think maybe not everyone knows it, but Sam is incredibly passionate about this subject. It would be great to hear..."}, {"start": 1397.86, "end": 1425.07, "text": "Why you're passionate about it, but also just to talk about what's so special about this AI track and what type of support you're providing to these founders that's quite unique. So about a year and a quarter ago, a group of us started, it was the first project that YC Research ever funded, this group called OpenAI. And at the time I had, and I studied AI as an undergrad, and at the time I had an inclination that things were about to go very fast. And now..."}, {"start": 1425.36, "end": 1456.22, "text": "With the benefit of hindsight, I knew I wasn't optimistic enough and that the progress that we're seeing in machine learning month over month continues to astound me. And I think anyone not paying attention is about to miss the most important technological shift of this generation. I think OpenAI will do great work. We are trying to sort of solve general intelligence. We're trying to build the algorithm for a program that is truly smarter than a human in every way and then figure out how to have that"}, {"start": 1456.22, "end": 1484.96, "text": "Maximally benefit humanity. And that's why we're a nonprofit. We don't ever want to be making decisions to benefit shareholders. The only people we want to be accountable to is humanity as a whole and doing this thing right. This is this mega transition from biological to digital intelligence, from humans being the dominant species in the planet or the universe to not being. And we want that to go well."}, {"start": 1484.96, "end": 1514.64, "text": "Sort of high stakes. It's a little bit stressful. So that is one effort. And that is very much a pure research nonprofit. Separately, we realized that in the shorter term, machine learning is going to remake basically every vertical. And that you're already seeing this with self-driving cars and AI radiologists that are better than human radiologists."}, {"start": 1514.64, "end": 1544.62, "text": "And AI lawyers that are better than human lawyers. And basically, I think, all repetitive work. We have the technology today, to a first approximation, to do all repetitive work, which is probably 75%, 80% of what any of us do. We haven't figured out creativity. We haven't figured out curiosity. And I think OpenAI will get that at some point. But today, we can do this incredible thing that seemed impossible just five or 10 years ago."}, {"start": 1544.62, "end": 1572.56, "text": "Three years ago. And it's such an important transition that we thought for the first time ever we would do a vertical and say, you know what? These startups, we want to offer special things to them. We want to put them together. We want to create a sub-community for them. Because they need some different things. And so we got special compute. We got access to special experts to advise them. And I think this is..."}, {"start": 1574.03, "end": 1605.22, "text": "I think this is something we'll do more of in other areas when they become this important. But even if not, I think this will be an interesting experiment. So you're going to provide people like Andrew Ng and other folks like that. Maybe not Andrew, but very, very good people to help in a specific domain area because you think it's the most important thing that YC could be working on. Again, I think there are a lot of important things. And one of the nice things about YC is that because we fund startups,"}, {"start": 1605.22, "end": 1633.74, "text": "Unlike companies, product companies, which end up in practice doing one thing really well and other things not as well usually, we can do a lot of things at once. So I think there are other areas that are hugely important. I think clean energy and the environmental impacts and also just the quality of life impacts is one that I am particularly passionate about. I think we are not that far from being able to cure many different kinds of disease."}, {"start": 1633.74, "end": 1663.31, "text": "A lot of things really important to the world that we have to do all at once. We'll come on to talk about some of those. But AI just feels like it's this wave that's about to crash over every industry. And there's so many companies there that we wanted to fund that we thought it'd be good to do something all together. One thing that you called out in your announcement about this was about robot factories. And you and I were at an event a couple of weeks ago. We saw a robot factory for replacing a salad bar."}, {"start": 1664.02, "end": 1691.31, "text": "Manufacturing line. What are the coolest robot factories that you've seen so far or you think are going to change the world? Again, I expect that almost every job in a factory will be replaced by a robot in the next, let's say, 10 years. And there are mega consequences to this. This is huge disruption and job elimination in a way that is scary."}, {"start": 1691.31, "end": 1721.2, "text": "On the other hand, I don't think people working on an assembly line have found their highest, most fulfilling calling. And I take the mindset of we owe it to society as a whole to retrain people whose jobs we displace. But if we can make them happier and have a better job and put their efforts, their talents to better use, that's still a net win. And in the short term, as we quote-unquote retrain people or help them re-skill,"}, {"start": 1721.2, "end": 1751.54, "text": "What are your views about taxing AI, taxing robots, or finding other ways for the government to redistribute? How important is it for policymakers to get involved given the scale of the change that you've just described? The taxing robots thing sounds good, but I think it's a sort of silly thing once you think through all of the consequences of that. However, in terms of taxing wealth and redistributing it, I think we need to do a lot more of that. And that is going to just"}, {"start": 1752.05, "end": 1782.11, "text": "Be the future. I think technology is a lever on ability and a concentrator of wealth. You get more total wealth, but it's concentrated in fewer and fewer people's hands. I think we are not that far away from being able to eliminate poverty effectively worldwide, certainly in developed countries. I think we are not that far away from being able to offer universal health care to everyone. And I think there are things that are 50 years from now will be basic human rights."}, {"start": 1782.11, "end": 1811.06, "text": "Healthcare and enough food to eat and a place to live. And we will do that. I'm confident we're going to get that done. But it is going to take a change in how we think about redistribution. And it's going to be uncomfortable to people because even though it's nothing like socialism, it's going to feel like it. And I think the societal compact we will have is some version of a floor and no ceiling. So there will be trillionaires, but they will pay a ton of tax and there will be no poverty either."}, {"start": 1811.09, "end": 1840.59, "text": "So it's still going to feel, I think, really unfair and like there's a huge delta. But my guess is that the societal trade-off will be people get to get really rich and there's an incentive to get really rich, but no one is really poor. So it'll be basic universal income or something like that. We're running a test on that. I was more skeptical of it initially. I'm now more optimistic that there exists a version of basic income where people still are..."}, {"start": 1840.88, "end": 1872.05, "text": "Sam, you're passionate around AI. Maybe you could just talk a little bit about what you thought pretty deeply around how this all plays out. I've seen you talk about the human-AI interaction and how you..."}, {"start": 1872.05, "end": 1903.28, "text": "Sort of see humans coexisting along with these general AIs that you're researching. Can you just sort of paint a picture for all of us and also maybe attach a time frame to it? Well, I can paint a picture. I can't attach a time frame. I believe in one or the other. I am always open to a better idea. I'm always willing to change my mind with more data. As I currently see the world, I think the most likely good long-term future is some version of Emerge."}, {"start": 1903.28, "end": 1931.94, "text": "I think if you have the AIs and the humans, and they're both trying to enslave the other, and they both want the same thing, which is sort of dominant species in the universe, that's bad. That's a recipe for conflict every time. We've played this experiment out many times in human history. When two people want the same thing, and there's only one of it, they're going to have conflict. And so most of the sci-fi novels are versions of this fight."}, {"start": 1931.94, "end": 1960.96, "text": "Man versus machine, one way or the other. And I think we should try to avoid that. I think, while I remain very open to being convinced otherwise, my current thought is that some version of a merge, and this can go anywhere from plugging electrodes into our brain and literally merging, or uploading our brains to a computer, or just every person has a little chatbot that becomes an extension of their will in the same way that..."}, {"start": 1960.96, "end": 1994.83, "text": "You know, our more complex brains are sort of in service of our lower brains. That might work as well. But I think we're going to need something like that so that we're one thing and it's not us versus the AI. Time frame, I try not to... You're never wrong if you don't guess at time frames. But in our lifetimes, we'll see it. Well, I think we're all going to live pretty long, too. But yes. Just moving on, you've, I think,"}, {"start": 1994.83, "end": 2027.76, "text": "You said publicly that you think that solving the world's access to cheap, abundant energy may actually solve almost all the world's problems. Could you maybe just talk about this? And also, I mean, you're obviously on the board of Euclid Fusion businesses, which sounds very exciting. Maybe you can share kind of what your vantage point is from those companies. I've studied the history of technology more than I've studied any other part of history. And one of the things that becomes very apparent"}, {"start": 2027.76, "end": 2058.37, "text": "Is that quality of life correlates with cost of energy. And every major breakthrough we have in terms of cost of energy leads to people's lives getting a whole lot better. And so to some sense, if you can just make energy ten times cheaper, you can make everyone's lives a lot, lot better. And then there are all these other problems related to energy. Absolute destruction of the environment. The cause of most wars fought in the last several decades or longer."}, {"start": 2058.37, "end": 2088.62, "text": "And I think the coming needs for huge amounts of energy for AI supercomputers. So I think it would be a really good thing if we could get to huge amounts of very cheap energy. And I think the two things that will get us there, have a chance to get us there, are nuclear and then solar plus storage. And I think we should pursue both. Probably one will be the 80%."}, {"start": 2088.62, "end": 2118.19, "text": "And the other will be 20. But I think it's really important that we get there somehow. A number of people studying this have felt like the next big wars are going to be fought over access to water. And with desalination technology and the energy intensity there, this could solve those problems. Yeah, again, all these problems reduce often to energy problems. So if you can solve, if you can make..."}, {"start": 2118.19, "end": 2150.0, "text": "Sufficiently cheap entity and sufficient quantities. You can go a long way to address the food and water problems as well. Maybe we could just keep going and just talk about a few of the other big tech trends. You're also quite involved in some of the work around synthetic biology. Ginkgo Bioworks is one of the well-known companies that you're involved with. What are some of the things you see on the horizon around that and the whole bioinformatics space? I don't think it's hit most people. I don't even think I've fully internalized the fact."}, {"start": 2150.0, "end": 2180.35, "text": "That we can now edit DNA. And sure, we can't do it perfectly and only in limited cases, but we can do it now. And, you know, this is like a program that's now able to edit itself. And I don't quite think I know everything that that, all the consequences of that. But I know they're going to be big ones. And there are the, again, the sci-fi novel dystopias of people can now engineer."}, {"start": 2180.35, "end": 2207.47, "text": "Really terrible synthetic pathogens. We can also hopefully engineer better medicines to treat those. And we can edit our future offspring. We can edit single cell organisms to produce anything we like. That's what Ginkgo Bioworks does. They can make you a molecule, a difficult to make molecule, by creating a gene and putting it into a bacteria and then growing a lot of that bacteria."}, {"start": 2207.57, "end": 2238.66, "text": "We can edit life. That's just an odd concept. And it's one of these things that went from being science fiction to possible more quickly than our regulatory system can react. And I think there will be huge upsides to that. I think there will be some real downsides as well. But it's still an area that I think the startup community doesn't pay nearly enough attention to. So we're going to open it up for questions just after this question. So maybe if you guys have questions, there will be some handheld mics that are going to be..."}, {"start": 2238.66, "end": 2268.56, "text": "And there's a mic at the front as well. So let's do just one last question from our conversation, then we'll open it up to the floor. U.S. politics. We don't have time to really talk much about this, but you're seen at the San Francisco airport expressing your discontent with the Muslim visa bans."}, {"start": 2268.56, "end": 2298.4, "text": "You've obviously been very involved in a number of initiatives, including, I think, a new GovTech initiative that you're working on at YC. I think just from this international audience, it might be interesting to hear what your concerns are about the shift in the US's approach to the world community and whether or not the tech community is doing enough about that. Look, I've never felt so politically motivated in my life. I think that what's happening in the US right now"}, {"start": 2298.4, "end": 2331.66, "text": "It scares me a lot. I understand some of the reasons why now in a way that I didn't earlier. I still think what's happening is really bad. I have to say I've been to a lot of countries and a lot of people make fun of me for US politics. I didn't expect it from Brits. But I think it's still a fair point. And I think these are trends that we're going to see a lot more of."}, {"start": 2331.66, "end": 2360.21, "text": "If we can't figure out what tech can do to make the future good for everyone, not the small set of people that run startups. And I think also, tech has just absolutely enormous influence in our lives. I talked earlier about the merge. Maybe it's already started. I feel very, I don't know about you all, but I feel very controlled by my phone. And although it's not technically part of my hand,"}, {"start": 2360.21, "end": 2390.37, "text": "This hour that I'm not holding it is uncomfortable to me. And I do what it tells me to do. It buzzes and I go do something. It shows me something and I feel happy. It shows me something else and I feel sad. And the algorithms that decide what I get to see are, again, not physically part of me, but in a very real way they are part of me. And we have a huge responsibility as tech companies to think about that. The things that make a company's graphs go up and to the right, the things that incite those passions in people,"}, {"start": 2390.37, "end": 2420.18, "text": "Are not necessarily the things that are best for our society. And there's the fake news issue. There's the filter bubble issue. But we better get serious really soon as a tech industry about doing things that unite us. We have learned that we can do things that divide us and have huge user engagement. And that, you know, that is a settled question at this point. We can do that. But if we don't figure out quickly how to build technology that unites us,"}, {"start": 2420.18, "end": 2448.77, "text": "If we don't figure out how to make everybody win, if we don't figure out how to make everybody not just win economically, but have a say in the future and control and autonomy over their lives, I think we'll be in a really bad place. And I think Trump in the US is a shot across the bow to that. Trump was a protest vote. People were saying, you know what? You never listen to me. This time you're going to listen to me. And I hope it is a short black mark in our history."}, {"start": 2448.77, "end": 2477.2, "text": "That yields a great response and is a rallying cry for people to come together. I am terrified of the alternative, so I am very motivated to help try to rally tech together to make sure we get the, you know, the future is always these branching probability paths. Really motivated to make sure we get the good one. So a number of CEOs in the tech industry have stood up to be counted."}, {"start": 2477.2, "end": 2506.75, "text": "And a number have not. And those that have not have argued that they have employees that support a certain direction. They have employees that support other directions. And so how do they really genuinely represent their employees? What's your advice to those founders? Look, I think there are a lot of good reasons to vote for Trump. I think there are probably a lot of good reasons to vote for Brexit. I think in this desire to have everything be on one side or the other,"}, {"start": 2506.75, "end": 2536.43, "text": "We lose sight of the fact that sometimes people do something we really disagree with for some good reasons and some bad ones. And the quote, I will work with anyone to do good and no one to do evil, I think is important. And people that choose to work with Trump to try to do as much good as possible, they have a stronger stomach than I do, but I'm sure glad they exist. I'm sure glad that Elon Musk is working with Trump. And I think I totally understand why founders or CEOs don't want to oppose Trump."}, {"start": 2536.43, "end": 2565.3, "text": "But I also don't think there's a conflict in between saying, I understand there may be very good reasons that you voted for Trump, and I support those. But I'm going to stand up against these policies that are so against our interests. So I think it is possible to say, and this is the position I've taken, I understand that people support Trump. I want to understand it more. I want to understand the good reasons and what I should update about my own worldview and where I should change my mind, but also"}, {"start": 2565.3, "end": 2595.7, "text": "When he tries to ban Muslims from our country, yeah, I'm going to go protest. And these aren't inconsistent to me because any politician, any political movement is a complex set of ideas, some of which I will agree with and some of which I won't. And I think most people understand that. It doesn't play as well for the soundbite, the tweet, social media posts, but it is the reality of how the world works. And I think it's important to respond issue by issue."}, {"start": 2595.7, "end": 2632.26, "text": "Rather than saying all Trump voters are terrible people, which I obviously don't believe. Do you think when you're programming the AI for the future US president that he'll be excluded from the data set? There's always this hard question about what you count as an outlier and what you don't. Maybe we should open it up to the floor. Does anyone have a question? I think we've got sort of roughly..."}, {"start": 2632.26, "end": 2685.23, "text": "10 or 15 minutes for Q&A. If you could try and make it short and crisp, that would be great. I'll repeat the question. We'll repeat it for you. Go for it. The question is, AI is hard and startups are hard. Are they somehow synergistically together too hard? I don't think so. I think AI is not... Applying existing"}, {"start": 2685.23, "end": 2714.82, "text": "AI technology is not as hard as people think, and there's always something that's hard about a startup. A good startup always requires exactly one miracle. We are going to have this special track, and we will try to make available the tool sets, the resources, the compute, the data that AI companies need. But I don't think it's harder than a lot of other things. And you also have this huge tailwind of it's this trend a lot of really smart people want to get involved in."}, {"start": 2714.82, "end": 2752.62, "text": "I wouldn't worry about that. Luckily, I'm very loud, so I'll just give that a try. Sam, first of all, thanks for a brilliant talk. And I know I've been looking at startup school and what you guys are doing with that, and I really believe it could be kind of the alter basic moment for startups around the world to really scale up and really motivate people and believe that it's possible at a truly global scale. So I really applaud what you've done there."}, {"start": 2752.62, "end": 2781.31, "text": "Thank you for that to begin with. And the reason I think that is because when Startup School first opened, I looked at it with my co-founder. We thought, this would be amazing for us. This would be brilliant. And then we very quickly thought, well, hang on, if it's amazing for us, it'll probably be good for other startup founders in London too. And I know a big YC belief is, well, go out and find your users. Find something that people want to do. So I signed up to do nine meetups in 12 days."}, {"start": 2781.31, "end": 2807.68, "text": "I went out to meet people, to meet founders. Can we get to the question? I'll get to the question. Very rude. I went to meet founders and so far we found 26 of them who are going to be working with us on what we're calling YC London. And some of them are here today. Some of our founders."}, {"start": 2807.68, "end": 2831.06, "text": "And I know it's all about building something users want. So here's the point. There is a point. The point is I spoke to our founders, our users. What did they want? I know you're a big believer in ask for things that you want. So that's what I'm going to do. The big thing we want, we're doing YC dinners, one a week. We've got speakers coming along to them. The big thing our founders want is to have some London YC alumni come along to our dinners."}, {"start": 2831.06, "end": 2857.23, "text": "And be able to speak to our founders and offer some advice, offer that learning. I know some of them may be in the room. Well, I think some of them are in the room, so you should corner them after and make them come speak. Do you think that would be, they should come and speak to us and that would be a wonderful thing? Sure, come speak to you and maybe you can get them to speak. Good luck. Thank you very much. Sure. Hi, this is a question."}, {"start": 2858.26, "end": 2887.57, "text": "So with AI and robotics going to take a lot of jobs across different industries like retail, manufacturing, millions of jobs, what do you think tech startups can do from a very practical point of view in the job retraining and education space? Look, I think, great question. You want to start a startup where you have a wave coming bigger than you that you can leverage, and you are guaranteed one."}, {"start": 2887.57, "end": 2917.33, "text": "With the need for retraining. So we're actually going to put out a request for startups, specifically for people that want to offer job retraining. Because we know there's a huge market there, and it's a really important societal question. So I think it's the right market to go after. In terms of the way to do it, there are a lot of different ways. I probably can't give any kind of valuable one-minute answer. But I would study how the effect, and I have been studying this, the effect of adult education programs in the past."}, {"start": 2917.33, "end": 2946.35, "text": "And I would look at what's worked and how it's worked. And I think the more I've read about it, the more optimistic I've been. So I think it's fertile ground. Good luck. Thank you. So YC doesn't really offer any lab or office space. But for a biotech company, it just seems like some of us need that. So what is YC to do? Well, we do have a deal for the biotech companies to get lab space."}, {"start": 2946.35, "end": 2974.62, "text": "And we understand people need that. So we don't offer a general office space for our startups because we think that's really bad. We think that you want to be alone just for your own startup. You need your own door, your own identity. And you need to be alone mostly so that people can't tell you why you're going to fail, which I think is the thing that always goes wrong in co-working spaces is that it's just constant negative energy. But for biotech startups, for hardware startups,"}, {"start": 2974.62, "end": 3008.56, "text": "Although we don't have a big open plan shared office, we have a number of deals set up where you can get free access to hopefully anything you would need. Next generation sequencing? Yeah. We at one point had sequencing machines no one else had. Now probably they've caught up. But yes. Hey, thank you for your talk. Just a quick question on blockchain. What do you see as the most immediate opportunity?"}, {"start": 3009.68, "end": 3036.38, "text": "Look, I think Bitcoin should win. Although the community is so toxic and so dysfunctional, maybe it won't. But historically, anything that gets this big, because network effects are so powerful and order N squared working the way it does, it should be done. It still should win, even with this really dysfunctional community. And so I think this idea that Bitcoin is weak and I'm going to create a bunch of altcoins is probably wrong."}, {"start": 3036.38, "end": 3066.85, "text": "Maybe Ethereum gets escape velocity, but even that I would still kind of doubt. I think instead of saying I'm going to create a better version of this, it's better to say, well, this will probably be the platform that sticks in the same way that SMTP was broken but stuck, and we are going to focus on building new applications on this. My sense is that's better, a better use of time. And I think within that very broad framework, there are so many good ideas."}, {"start": 3066.85, "end": 3100.98, "text": "that range from smart contracts to proof of identity to everything else that it's kind of, again, hard to give a woman an answer. It would sort of depend what you're interested in and what you're good at and what your background is, but I would generally look in that direction. Hi, my name is Sachini, and I feel very privileged to be here. So thank you. Can you hear me? I'm interested to hear a bit more in how technology might be able to help with the redistribution of wealth, and you started..."}, {"start": 3101.17, "end": 3130.96, "text": "I think there's two general schools of thought here. One is that you should just tax and redistribute and everyone as a human right should get enough money to live on and have an opportunity to learn new skills and do whatever they want. And the other is that you should continue to trust capitalism and just accept that we're going through one of these transitions."}, {"start": 3130.96, "end": 3156.4, "text": "On the order of the industrial revolution or the agricultural revolution, that doesn't happen very often. And, you know, there's plenty of jobs on the other side because humans figure out new things to do. Human demand is unlimited and we'll always want new things. The reason why it might be the latter is that every time we think this time it's different, it's usually not. And sure, computers are different than..."}, {"start": 3156.88, "end": 3184.29, "text": "You know, factories, but maybe not that different. And probably we'll find all sorts of new stupid things to want and compete over and new ways to create status. And I think that's probably right. And so I think capitalism probably continues to work fine, although it's going to be a rocky transition. The other point of view, of course, is that this is the last technology revolution. And that AI, because it's better than humans at everything, or could be better than humans at everything, the world is really different. In which case..."}, {"start": 3184.29, "end": 3211.63, "text": "You do need something super different than what we've tried before. And I'm certainly not smart enough to know which of those two it's going to be. In any case, I think it's good to be thinking about both. Redistribution and retraining, I think, should both be kind of major societal priorities. If we do end up in the redistribution path, if humans aren't better than computers at enough stuff, I think the one thing that's really important and we're trying to study in our basic income study..."}, {"start": 3212.14, "end": 3242.19, "text": "Is if people don't get the meaning and the social environment and the fulfillment and the respect that they normally get from their job, what new systems do we have to create for people to get that? And that, to me, is a much more interesting question. Then will we have enough money to afford this? Because, of course, if machines come and do all the jobs for free, the cost of everything is going to go way down. We'll be able to afford it easily. I have no interest in that question. But what society looks like, that's a hard one. Thank you."}, {"start": 3242.19, "end": 3272.77, "text": "Sure. So I have a little philosophical question. It's a two-part question. First is that, how do you measure your personal success? And second is that, given that in the current state, organizations seem to have a purpose of enhancing shareholder value. But for a person, money is not everything. So how do you incorporate or design that systematically into the organizations? I didn't quite understand the first question. How do you do what success? How do you measure your personal success?"}, {"start": 3272.77, "end": 3299.07, "text": "Measure persistent success? Your personal success. Oh, my personal success. You know, I think like a lot of other people, when I was just starting out in my career, I was focused on making money, or at least enough money that I could do whatever I wanted. And the moment I crossed that threshold, I honestly don't think about it much anymore. And at this point, it's like, do I wake up every day and do something I want to do with people I want to do it with?"}, {"start": 3299.07, "end": 3327.68, "text": "That I believe is somehow giving back to the world and making it better in some way where I think I can uniquely help. The times I'm happiest are when I'm doing something where I think if I don't do this thing, it's not going to happen. And it's important. And so for me, success is if I wake up most days and I feel that way. And right now I do. And if I stop, one thing I'm really good at, which is a skill that I recommend cultivating, is as soon as I don't feel that way, I stop what I'm doing and do something else."}, {"start": 3327.68, "end": 3355.89, "text": "And so like, you know, like my partners kind of joke with me that I have a short attention span and that I just like wake up one morning and decide to hand over some big piece of YC to somebody else. But I do that so that I can work on new things once I feel like I've kind of gotten something in motion. I, you know, the thing that I've turned out to be good at is getting things started and sort of figuring out what needs to get started. And I also like that."}, {"start": 3355.89, "end": 3382.54, "text": "So I try to optimize my time to focus on that and hand over the other stuff to other people. The second part of the question was that how do you incorporate that sort of fulfillment into the organizations? Because the organization seemed to be just about making money, but you as a person are seeking sort of fulfillment and sort of making sure that some objective is fulfilled. How do you incorporate that systematically if you had to think about it?"}, {"start": 3382.54, "end": 3410.03, "text": "You know, most organizations tend to be reflections of the people that run them, their founders. And so I don't... That kind of happens. I think in the same way that children are somewhat like their parents. That kind of happens without me trying. I think people that want to work with me kind of know what I'm like. And it doesn't work for a lot of people. It works for some. And they know what I care about. And the organizations that I... Or the things I spend a lot of time on tend to reflect my values."}, {"start": 3410.03, "end": 3456.72, "text": "And that happens without a big effort. I think the more you're just open with people about what you care about, what you want to do, what you're motivated by, the more that takes care of itself. Thank you. Sure. One question. Yeah. Hi. We've mentioned the need for tech to unite all of us in close up. This is here in London. I'll repeat it. Go ahead. For diversity."}, {"start": 3456.72, "end": 3486.16, "text": "Sure. Current state of diversity in tech and how Y Combinator helps it. Well, look, I always think that if we can just do something rather than talk about it, we should do that. So, you know, we have more women in our partnership and more than the top 25 venture firms in Silicon Valley put together. We fund more women, we fund more black people, we fund more Latino people than the top 25 venture firms put together."}, {"start": 3486.16, "end": 3518.74, "text": "Our core program is run by a black guy. Another one of our programs is run by a Muslim guy. I'm gay. We have a very diverse partnership. And I think that's reflected in the founders that we fund. And I think rather than talk about it a lot, we just try to live it. And we've been able to have what I hope has been a big impact there. And we'll keep doing it. And I think that's the thing other people can do as well. I think it's great to talk about it. It's great to bring attention to it. But it's much better to just do it. We've only got time for a couple more questions."}, {"start": 3520.3, "end": 3560.85, "text": "Which public decision-making process do I think will be most successful now or in the future, and would I ever give my vote to an AI? On the first one, I don't feel qualified to answer. I don't really know. I've studied this, but I haven't formed a conclusion that I have confidence is correct. Would I give my vote to an AI? Yeah. I mean, I would need to understand it and see evidence of it over a long period of time, but sure."}, {"start": 3562.9, "end": 3603.6, "text": "I'm very willing to admit that I have all kinds of bugs in my brain and I'm not very logical and I don't think any of us are. And one of the things that's exciting about AI is that we will be able to edit the bugs out hopefully. How do you know when to hold on to an idea that's failing or let it go and pivot? It is hard to answer in the general case. I can usually answer it if I sit down with you for 20 minutes."}, {"start": 3603.6, "end": 3632.16, "text": "The two general frameworks that I would think about, one is when you're out of ideas. If it's failing and you have more ideas, you can keep trying. If it's failing and you're out of any ideas that you really believe in, that's usually a good sign about when it's time to give up. And the other is if you have a few people that really love what you're doing, I think it may be good to keep going. Ask them where you can find more. If instead you have a lot of people, but they only like what you're doing a little bit,"}, {"start": 3632.16, "end": 3659.94, "text": "I'd be more likely to give up and try something new. Sure. I think we should probably make this the last question. I apologize to all the folks standing in line patiently. Unless it's a very, very good one. Sure. This is a really brief one. You spoke briefly about the characteristics. So people that have come in as founders have been very determined. But can you speak a little bit about the behaviors that you've seen from people?"}, {"start": 3659.94, "end": 3693.94, "text": "And the behaviours that you've done is to make successful founders. The behaviours that make successful founders? Yeah, as opposed to the type of domination. Yeah, the question of work-life balance is very difficult. I think it works to have a startup and one other mega priority in your life, and then some other medium-sized ones. But I always try to be open with people that if you choose this path, it means sacrificing a lot of other areas."}, {"start": 3693.94, "end": 3723.5, "text": "And that's just, that's hard. It's fulfilling in its own way, but it means sacrificing a lot of other areas. But no balance at all usually leads to burnout as well. So I think everyone finds balance in their own way, and it's kind of this oscillating function that's never quite perfectly in tune, but eventually gets closer to converging. In terms of characteristics, I think one thing that's important to learn, even if you're not naturally good at it, it's learnable, is being a good communicator."}, {"start": 3723.5, "end": 3752.43, "text": "I think I may have mentioned that as a characteristic, but for me and for most other people, it's a behavior. I'm not naturally a good communicator. I'm probably not even with practice a good communicator, but I have gotten a lot better. And the ability to clearly communicate ideas is so important and so undervalued that if I were going to say, here's one thing a founder can practice and learn and get better at, that would be high on my list. The other thing I think you can learn to get better at"}, {"start": 3752.43, "end": 3776.24, "text": "It's how to have unpopular ideas. Most people don't have their own ideas. Most people are a regurgitation of other ideas they've heard. And I myself used to be like that. And I realized that I could make a mental effort and I could practice having my own ideas and not just being influenced by whatever the last reasonably smart person I spoke to talked about."}, {"start": 3776.56, "end": 3799.52, "text": "Now I don't have to make a conscious effort about that. It just happens. It's a more interesting way to live and think and definitely more effective for what I do. But I think that is a learnable behavior, but it takes significant effort. Sure. No pressure, but the very last question. So I'm using an AI coach, which is to be the best coach in the world."}, {"start": 3799.52, "end": 3810.83, "text": "And I have two questions very briefly. One is, I would love to show you the product because I applied for Y Combinator. That's not a question."}, {"start": 3811.09, "end": 3838.96, "text": "The question is about vertical versus, you know, the big market. So I keep on getting, you know, different advice on should I create the best coach in the world by looking at the whole wellness area or should I just focus on one thing? Sure. All right. So first of all, an advice about advice or taking advice. You will often hear a lot of different opinions. First of all, almost no one knows what they're talking about. And second, there's often multiple ways to do things."}, {"start": 3838.96, "end": 3867.95, "text": "The only thing that is disastrous is to let yourself continually get slingshotted around from one thing to the other. What you want to do is listen to a lot of advice, internalize it all, think about how it might fit together, and then make your own decision. This is what you are paid to do as a founder, is listen to a lot of advice, realize most of it is bad, integrate the different pieces that are good, and then do that. And once you've committed to that, don't let other people who probably don't know any more than you do keep telling you that you're screwing it up."}, {"start": 3868.02, "end": 3898.22, "text": "For this specific thing, though, I have a high confidence piece of advice. What you want to do is start with a narrow vertical, dominate a small market, and then expand from there. The two choices usually you face are conquer a small market and then expand, or start in a big market with a small market share. And having watched more than 1,000 startups with high confidence, dominate the vertical first and then expand. Sure. Thank you very much."}, {"start": 3909.52, "end": 3939.98, "text": "Just before everybody gets up and leaves, I just want to thank, first of all, Shortage Town Hall for hosting us. Unfortunately, we do have to wrap up things now so that they can clear the venue. But I want to thank Sam Altman for his fantastic contribution. Thank you all for coming. We need a much better representation from London applying to YC and getting into YC. So it's all on you guys now. Thank you. Thank you."}]}