Astro Teller: "Radical innovation almost always doesn't work. Anyone who pretends differently is telling you a just-so story after the fact."

Ilana Golan, host of the Leap Academy Podcast: "Dr. Astro Teller is the co-founder and captain of Moonshots, the CEO of Google X and Alphabet's Moonshot Factory. He leads the creation of breakthroughs, technologies, and innovations like Waymo. We're going to geek this out a little bit."

Astro Teller: "The breakthrough technology that gives us a sense that there's at least a glimmer of hope."

Astro Teller: "In the early days, we were having Googlers who used them for their commute. We said, 'Hands right by the steering wheel. We're going to have cameras in the cars looking at you.' We saw people getting into the back seat, doing their makeup. One of them fell asleep."

Astro Teller: "It was never our job to make and sell a car. Our job is to make the world's best driver."

Ilana Golan, host of the Leap Academy Podcast: "This is fascinating. I could probably talk to you for hours."

Ilana Golan, host of the Leap Academy Podcast: "Dr. Astro Teller is the co-founder and captain of Moonshots, the CEO of Google X and Alphabet's Moonshot Factory, where he leads the creation of breakthroughs, technologies, and innovations like Waymo, the self-driving cars — which, by the way, we're the biggest fans of, just FYI — Google Brain, and Wing, which is redefining how bold ideas become real-world impact. So cool. But before that, he was also a serial entrepreneur, a scientist, and a former Stanford professor. Astro has built and exited multiple companies, holds numerous patents, and is recognized as one of the leading voices in innovation and future thinking overall. It's just incredible that we get to hear him on this show. Thank you, Astro, for coming here."

Astro Teller: "Ilana, thank you for having me. *[laughter]* I've really been looking forward to this conversation. It's going to be really fun."

Ilana Golan, host of the Leap Academy Podcast: "And for everybody here listening, this is going to be a fascinating conversation about how, first of all, he got into Google X, but also how they created not only real self-driving cars, finally, but also — we'll talk about drones and AI and the future and so much more. And also, remember: once a week we choose a question from our YouTube channel, and in Leap Academy with Ilana Golan I answer it here live at the end of the conversation. Today I actually want to invite Astro to answer this with me, because the question from Greg was, 'What do you think is around the corner that we are not aware of?' And I love that question. So, Astro, let's do this together."

Astro Teller: "Let's do it."

Ilana Golan: "Okay. But first, I'm going to have to take you back in time a little bit. With all your extensive career and everything you've achieved, can you give us a really quick recap? How did we even get into Google? Why does Sergey come to Astro and say, 'Come to Google'? What happens there?"

Astro Teller: "I'll give you a few different vectors in. One of the ways to think about it is that Google is known for giving people the opportunity to work on 20%-time things. Inside Google — now Alphabet — they've talked for a long time about 70% being the core work, 20% being the adjacent work, and then there's 10%, which is the really-much-farther-out-there kind of stuff. And they had a position open for quite a while called the Director of Other, which was meant to take care of that 10%. They interviewed me, and I think several other people, and didn't fill the role. But Larry and Sergey had been thinking for a long time, before Alphabet was created, about the fact that there's a physics to businesses: as they get larger, they get more complicated. It's nobody's fault. It's just the physics of businesses. And they wanted to make sure that what they imagined as Alphabet — even before it had that name — would be able to keep stretching itself outside its comfort zone. And so X, the Moonshot Factory, was set up 16 years ago with that as our mission: don't solve Google's current problems. Google's good at solving Google's current problems. Go find big new problems in the world that we could be really proud of solving, and then hopefully find some world-changing solution to those problems. So that's been our mission for a long time — to help Alphabet, even before it had a name, get more surface area with the world, and to create a series of businesses, which we call moonshots, that were particularly audacious both in the way they could help the world if they succeeded and in the quality of the business created."

Ilana Golan: "And maybe define 'moonshot' for the people listening."

Astro Teller: "So, what we call a moonshot has to have three components. The first is that there has to be a huge problem with the world that you can name and you want to solve — otherwise it's a bit of an academic exercise. Then, tell me a story. Tell us a story about some science-fiction-sounding product or service. Don't worry yet about whether you can make it, whether it's likely, or even how you would make it. But what's that product or service that we could pre-agree, if you could make it, would resolve that huge problem with the world? And then, what's the breakthrough technology that gives us a sense that there's at least a glimmer of hope in the early days that we could actually make that science-fiction-sounding product or service, so that we could resolve that huge problem with the world? And if we have that, we're not done. That's a moonshot story hypothesis. That's like opening up the starting gates. From there, great — that's a testable hypothesis. Let's go get some evidence that teaches us either that things are a little bit more exciting than we thought, or a little bit less exciting. If they're less exciting, cool, we'll stop it. If they're more exciting, great, we'll double down on it. That is the basic internal function of X. Back to your question about how I got into this position: X and Waymo came into being at about the same time, and at least partly, X came into being as a place to house this thing that ultimately became Waymo. As part of that getting formed, the founders of Alphabet asked a man named Sebastian Thrun to set up X, and then asked him if he wanted a co-founder. He came and asked me if I would do that with him. So, very practically, that's how I ended up here. For about two years, I was his first mate — we co-founded it, but I reported to him — and then he ended up leaving X after about two years to set up Udacity on the outside. At that point, I became the captain of Moonshots instead of the first mate."

Ilana Golan: "Oh, I love this story. For those who don't know Sebastian — correct me if I'm wrong, Astro — he's kind of the godfather of open online courses. I think that's completely disrupting education, and we're going to talk about MasterClass in a second, but it created a huge impact on its own. Okay, so somehow they come to you. But take me there for a second, Astro: how do you know if you're set up for success? If you just work on something very futuristic, how do you know if it's going to be successful? How do you even measure that? What does it look like?"

Astro Teller: "Well, spoiler alert: radical innovation almost always doesn't work. Anyone who pretends differently is telling you *[laughter]* a just-so story after the fact. That's just survivor bias. You cannot do radical innovation without being mostly wrong. So the main thing that happened while I was interviewing with Alphabet is that I kept asking, 'What is your tolerance for stopping things?' Because that is the rate-limiting step of innovation. If you don't have tolerance for that, I can't succeed here. And essentially everybody said, 'We want to be great at that, but we aren't yet great at that.' And I believed them — they at least wanted to be. So my only ask, or test, before I joined Alphabet to help co-found what is now the Moonshot Factory, was operational and cultural separation. Not because Google is anything other than great — Google's a really amazing company — but we needed to be, and to work, differently in order to produce something that would be different. I was convinced that in order to create an environment where it was rational locally to actually kill your own projects — to be in this much more rapid discovery and intellectual-honesty process — we needed a subtly but profoundly different culture. And that couldn't happen without the cultural and operational separation, which Alphabet and Google gave to me and Sebastian, which is why I joined."

Ilana Golan: "Right, and that's incredible. Google was always known for its innovation — like you said, the 10%, working on the future. But it's something very different knowing that you're going to need to kill your own projects. So how do you motivate a team, Astro? How do you create a team where the morale is, 'Hey, this is so cool, we were really passionate about this' — and now we're going to kill it? How do you not break the team?"

Astro Teller: "I'm going to play a game with you where we pretend that you're coming to join X. Don't worry, I'm not actually stealing you — but for this conversation, imagine you're considering coming to X, and you're asking me that question, which is the central thing you should care about, and I had better be able to answer it. On the surface, all we're asking for is intellectual honesty. Nothing more. Surely we can agree that if the thing you're working on is not actually the best way for X and Alphabet to spend its money trying to help the world — if God came and told us that was the truth — we could agree we should stop doing that thing. We both want to be intellectually honest. The trick is the way the world actually functions. Even though you want to be intellectually honest, and you would step away, even with some sadness, from something you'd worked hard on if it was no longer really promising — the world has taught you not to do that. That's a career-limiting move everywhere else. So the real question you're asking me is not whether I'm serious about it. It's: what are you doing at X? What are the hundreds of things you're doing at X so that it's not a career-limiting move to say, 'Hey, the teleporter I've been working on really hard for two years isn't working great. I'm not sure we're going to get there. I can't prove it won't work, but I think we should stop this'? So there are all these questions: if I do that, can I still get promoted? If I'm a really great Xer, can I get a bonus at the end of the year? How will I be treated by my co-workers at our all-hands? If I stand up and say, 'I've killed my project,' and everybody boos, not only will I never do that again, but nobody's going to do it again. If you get a standing ovation for killing your own project and for demonstrating your intellectual honesty, then you'll say, 'Oh, maybe it's okay to do that here.' And other people sitting in the stands will think, 'Oh, maybe I can be intellectually honest.' Which is why we work so hard to make sure that when people kill their project, we give them a standing ovation. And no one of those things can undo it. You have been taught since you were five — we have all been taught since we were five years old — to play the short game, to get an A+ on the test, to not say you don't know the answer. All of these are exactly the opposite behaviors from what would actually drive radical innovation, but that's adaptive, sadly perhaps, in our world. So X has worked really hard to create a microcosm where those reinforcements don't exist, where there's a different set of reinforcements toward the habits that are healthier for innovation."

Ilana Golan: "This is so fascinating, because as you were speaking — I started my career in the Air Force as an F-16 flight instructor. And one of the big things we would do there is be very harsh with ourselves. We'd have to stand in front of everybody and basically say all the mistakes we'd made and what the lessons were, every single time. There's a humbling thing about it — it feels like I'm naked right now — but that's what you need to do in order to learn really, really fast. So I love that you shared this. But I'll ask: we always train to minimize risk, and now you're asking people to train to 10x the growth, right? You can't do this if you think 1%. You have to think 10%."

Astro Teller: "Totally. Can I get on a tiny soapbox about risk for a second?"

Ilana Golan: "Yes, please."

Astro Teller: "You were absolutely right. The whole world is obsessed with getting risk as close to zero as possible, which is rational if the size of the prize doesn't really matter and you'll be punished if you don't succeed. That's why the whole world is so busy doing small, nearly meaningless, incremental steps most of the time — because they feel like they can't afford, as individuals and as subgroups, to not succeed. And once you have to get the risk to zero, you've basically thrown away almost all the upside you can get. So let's play a game: choice A or choice B. Choice A, you can give a million dollars of value to your business this year, guaranteed. Or choice B, you can give a billion dollars of value to your business this year, but it's not guaranteed — it's one chance in a hundred. So, choice A: a million, guaranteed. Choice B: a billion, one chance in 100. Which are you going to choose?"

Ilana Golan: "Duh. I'll choose B."

Astro Teller: "*[laughter]* You're going to choose B because B has 10 times the expected utility of choice A. It has an expected utility of 10 million; choice A has an expected utility of 1 million. That is a rational thing to do. What you've just done, by moving from thinking about risk by itself, is move to thinking about reward divided by risk. At X, I think everybody has internalized the idea that if you're working on something and you could find something with twice the risk but four times the upside, you should probably stop what you're currently doing and work on that other thing instead, because it has twice the expected utility. So moving people away from risk minimization to reward-to-risk-ratio maximization is one of the essences of innovation."

Ilana Golan: "So let me take you there for a second, because it sounds good on paper, but I bet there are some sleepless nights involved in this shiny world. Are there moments where you're like, 'Oh God, I don't know if this is going to work'? If you could take yourself back in time, can you share a moment where it's like, 'Geez, maybe we're too early for this kind of innovation'? Share a little bit, because it sounds amazing, but I'm sure emotionally there's a bigger roller coaster here."

Astro Teller: "There absolutely is. The tension is: on the one hand, you don't want to excuse-make your way into working on something that really is a zombie — where you keep it alive because 'just one more try, just one more try, you never know.' That would be bad; that's the opposite of what I'm describing. On the other hand, as you're pointing out, our most successful things — things like Waymo, things like Wing, and more recently our moonshot for the electric grid, called Tapestry, which is now profoundly important around the world for grid operators — I can't tell you the number of times we almost stopped them, because we couldn't figure out whether there was a there there, or which way we should go. What are we really doing here? Do we have something important enough to continue taking this kind of risk and spending significant money on? So it's an art as much as a science. That's just part of the reality: there isn't a right answer. Every one of these things is a bit of a snowflake, because you're always going — you're like an explorer going to a place no one's ever been before. You don't get a map. There isn't a checklist that teaches you to keep going until this point and then stop no matter what. It's a judgment call. And getting a group of people to agree is almost impossible. But there are lots of tricks — intellectual architectures — that you can put around people to help them think through this in various ways."

Ilana Golan: "So give us some tools. And by the way, I'll give you a little plug, because you actually talked about the grid innovation on your podcast — I was listening to it, and it sounds fascinating. But I do want to get some of these tools, and maybe we can also run through an example of Waymo, which started — what was it, 15 years ago, or whatever that adds up to? I'd love for you to take us back in time and look at what that looked like, because it's so easy to look at the cars driving now — they're so cool, and I've been one of your biggest sponsors, I love it, since we have it in Arizona and now in San Francisco. But the truth is, there were probably years of, 'Is this going to make it to market?' So can you take us back in time with Waymo?"

Astro Teller: "Sure. I'll give you two examples, exactly as you mentioned. For your listeners, if they'd like to, just search for 'Moonshot Factory' or 'Moonshot Podcast.' Our second season just came out a few days ago, and I think they'll really enjoy it. These are very unfiltered deep dives into the stories you and I are about to take little pinches of. But if people want to hear more, I think they'll really enjoy the podcast. So let me give you two stories — one from Wing and one from Waymo — since those are two your listeners will be familiar with. In the case of Waymo, we spent the first five years thinking we were going to make cars that drove themselves and sell them to people. That was kind of what cars were, so we were just going to make better cars. That was our conception. After about five years, we had them on the freeway, and we were having Googlers who didn't work at X take them and use them for their commute. We said, 'Okay, but you have to promise, promise, promise: keep your hands right by the steering wheel. We're going to have cameras in the cars looking at you. No messing around. The cars are really good, but this is public safety, so you have to keep your hands really close to the steering wheel all the time.' And these Googlers said, 'Oh, absolutely.' These Googlers are great human beings — way above average, I think, for diligence as drivers. And within days, we saw people getting into the back seat, eating their lunch, doing their makeup. One of them fell asleep. We had to stop the project, because we had been assuming that humans can be backups for technology that's not quite ready — and when the safety is this serious, that just doesn't work. Now, we could have stopped the project entirely — that was a big red flag relative to what we assumed our job was, what the business was. But we didn't. We had some serious conversations, and in the end we said, 'Oh, it was never our job to make and sell a car.' That was a misdescription of the mission. A real mission is to transform mobility — like an elevator: you should just get in, say where you want to go, and it takes you there. That's what these vehicles we're working on should be. We worked on that for another five years and came to a new conclusion, one that was much more business-related than safety-related: we still had it wrong. Mobility needs to be transformed, but that's not our job. Our job is to make the world's best driver. That's our job. There are a lot of ecosystem partners who can help us with the other parts, but what we should be great at is making the world's best driver. So there was real evolution in the thinking of what Waymo wanted to be when it grew up. And the first of those big evolution points was this crisis moment, when we realized that our whole conception of what would make it safe — that humans would be the backup — was a terrible assumption, and we had to abandon it."

Ilana Golan: "So it all starts — correct me if I'm wrong, Astro — with a fundamental mission, if you will. Millions of people die every year from car accidents. This doesn't make any sense. Robots can do better. Basically, it all starts with that, and then the question is which is the best vehicle to create that in. Am I reading this correctly?"

Astro Teller: "I certainly agree that the huge problem with the world is well described as: more than a million people a year die from car accidents caused by human error. There's also more than a trillion dollars a year wasted with people sitting in traffic unnecessarily, scraping up their cars, and a whole bunch of other stuff. Human lives are more important than the trillion dollars, but you can solve both at the same time, so that's a good thing. *[laughter]* The science-fiction-sounding product or service is a car that drives itself. You said 'the vehicle' — obviously the vehicles are evolving too, but I'd argue it's a complex mix of the vehicles, the sensors, the compute, the algorithms, and the data. There are a lot of different parts that go into making the world's safest driver. But over 16 years, we've gone from 'that sounds crazy' to it being in 10 cities, with lots of people getting into Waymos every day and having a ride — which is kind of mind-blowing for the first 30 seconds, and by the end people are just on their phones, back to doomscrolling, not even noticing there's *[laughter]* nobody in the front seat."

Astro Teller: "Can I give you a different kind of example, with Wing?"

Ilana Golan: "Sure."

Astro Teller: "Because we were talking about intellectual architecture that helps people past these crisis moments—"

Ilana Golan: "Yeah. And by the way, for those who don't know what Wing is, it's basically drones that deliver packages, right?"

Astro Teller: "Wing has been on a mission for about 13 years now to solve the last-mile problem. If you could just have anything that fits in a bread box appear anywhere you are — anything that could fit in that bread box, in a couple of minutes, for about a dollar — that's like nearly a teleporter. It would fundamentally change how we think, not just about things like food delivery. I have a hammer in my garage. The person next door to me has a hammer in their garage. We could probably share one hammer across 10,000 of us, but we don't have a way for the hammer to move around, right? Libraries are basically useless at this point — but if you could actually have the book fly to you, like Harry Potter Owl Post, and then fly home again, all of a sudden those books are not stranded assets anymore. There's just so much about the world this would change. So as we were coming to launch — this was about 10 years ago now — I sat down the whole Wing team and said: 'It is not today. It is three months from now, and our launch has been an unmitigated disaster. We feel physically sick. It went so badly we can't even look each other in the eyes when we walk by each other in the hallway. And you know why it went so bad. This is a test. You have two minutes. There's paper in front of each of you. Write down why it went so bad.' By moving people from pride, and the launch fever that can kick in, to that 16-year-old, 'I want to get an A on the test and prove how smart I am' kind of thing, everyone started frantically writing down all the things that were wrong with Wing. This was about a month before the launch. Once they'd written for two minutes, I said, 'Great. Let's organize everything you wrote down, sort the list, and start fixing things.' And we had a great launch."

Ilana Golan: "So good."

Astro Teller: "That's one of like a hundred tricks X has to help people work through — some people might call a softer version of that a pre-mortem. That was a very intense pre-mortem. *[laughter]*"

Ilana Golan: "I love that. That was so powerful — I could feel it in my bones. And I think you talk about a concept I really love. What's 'the monkey'? Can you share that? I find it so profound, because many times we look at the same thing but we fix the wrong problem, or we address something that isn't the riskiest assumption. Can you talk to us a little bit about it?"

Astro Teller: "Sure. In the early days of X, I kept saying — incorrectly assuming it would land well with people — that we should always work on the riskiest part of the problem first. In my mind, that sounded like 'duh.' But everyone at X clearly wasn't sure what Astro was saying — 'That doesn't sound right.' So I think I was having no effect on people, at least on that subject, for several years. And then I was at a Wall Street Journal Live conference, on stage, and someone was interviewing me, and I said, 'Obviously, you should work on the riskiest part of the problem first.' And the woman interviewing me said, 'That doesn't sound right. Can you explain that? Unpack that somehow.' In a moment of being hyperbolic, I said, 'Look, let's pretend we were working on a project and our mission was to get a monkey to stand on top of a 10-foot pedestal and recite Shakespeare. Which should we do first — build the pedestal, or spend our time training the monkey?' In that really extreme example, we know that if we built the pedestal first and said, 'Hey boss, look, I'm half done,' then in some really lame sense you're half done — but in a much more meaningful sense, you've burned down 0% of the risk. You've utterly wasted your company's money. The only thing that matters is: can you train the monkey? Because if you can't, thank God you didn't spend time building the pedestal. And if you can, we can get a pedestal afterward to put under the monkey. *[laughter]* Now, that's an extreme example, but in much more subtle ways, the world is awash with people who are building pedestals — not because they're stupid, and not because they're nefarious, but because their companies reward pedestals. They're just thinking, 'I want to feed my family. I want my career to go well. If I work on the monkey, it's probably not going to work — I kind of have a sense this monkey's probably not going to recite Shakespeare. I don't know what I'm going to do next, but I know if I build the pedestal, then I can say, "Hey boss, I'm half done," and I'll get a bonus, I'll get a promotion. And I can leave the monkey problem for whoever comes after me, or just hope the monkey thing doesn't come up until after I've quit and moved on to *[snorts]* another job.' *[laughter]* It sounds ridiculous, but this is most of corporate America. And this is another one of these intellectual tricks — finding a way to viscerally connect people to the important habits. We're trying to be like the card counters of innovation. Helping people learn the little tricks — and not just learn them intellectually, but really get them into their hearts and their guts, so they internalize them until it becomes muscle memory — is what it takes. Culture isn't what you put on the poster; culture is what people actually do. And that's why most of the work we do here is trying to connect people to these ideas in visceral ways that last and tend to get reinforced by our environment."

Ilana Golan: "Ah, I love that. There's a bunch of things coming to me, because one of the things we notice in Leap Academy — or any of these education companies — is that it's not so much about the content. Everybody runs to create the content, but the hardest thing is to create a client experience for someone who says, 'I'm stuck. I need to reinvent myself. I have no clue where I'm going next,' and they choose the Academy and actually get results. So I love the monkey example, because a lot of the time people are working on the wrong things — things that aren't going to help shape everything. I think one of my—"

Astro Teller: "Sorry — can I tell you one of our inside jokes, a recent one?"

Ilana Golan: "Yeah, go for it."

Astro Teller: "This is very X. A number of years ago, people decided our swag wasn't fun enough. So some group of Xers — I still don't really know who they are — started an Etsy store, which I think is now on Bonfire, called Swaggy McWaggerton, where you could get swag that's made by Xers but not approved by X. And at some point there was a joke here — I think you'll particularly appreciate this — that the really hard thing, the monkey, the hardest part of the problem, is actually us. It's an inside game. It's about managing our own psychology. So, to reference that the monkey is us, the monkey is inside us, there's now one of these designs where the monkey is kind of clawing its way out of the t-shirt that people wear around the office — to remind us that not only should we be working on the monkey, but the biggest monkey of all is managing our own psychology."

Ilana Golan: "But I want to tap into that, because I think mindset — especially in a moonshot — is going to be insanely important, right? It's this relentless, 'I'm not a failure. The fact that I haven't found this out yet is okay, because it's just going to take more time, more trials, more errors, more experiments.' How do you dial in that kind of mindset — a moonshot mindset, if you will?"

Astro Teller: "I'll give you an example or two, but no one thing will do it. You cannot just tell people, 'This is the right answer, go be like that.' You have to constantly reinforce it in all of these little ways. There's a flag behind me that says 'Keeping Alphabet Weird,' with a flying pig. I happen to be wearing the tarot card of the Fool. I don't believe tarot cards are magical, but *[laughter]* in all these different ways — I'm somewhat known for wearing rollerblades around the office — I want to remind people that we need to be in the beginner's mindset. In a way, that's the answer. But I just mentioned card counting a minute ago. Card counting is a process. Under the right circumstances, when you're playing blackjack and cards come across the table and are then retired, that slowly changes the statistics. Imagine the first four cards dealt were all aces — maybe a little odd, but it could happen. That would change the odds of how you should play the game afterward. If all the kings came out next, that would change the odds even faster. So people can actually get an edge over the casino playing blackjack if they do two things: they're very careful about watching what cards have come by, and they deeply understand the statistics and how they change depending on which cards have already gone across the table. Now imagine you claimed you could card count, and I was going to go to Vegas with you and give you my money—"

Ilana Golan: "Me too, me too."

Astro Teller: "I'm *[laughter]* not particularly good at card counting myself."

Ilana Golan: "Okay."

Astro Teller: "But if I gave you money because you claimed you could card count, and then you just went wild, randomly doing stuff, and you happened to win, I'd say, 'Shame on you,' and I wouldn't give you any more money — even though you'd won, even though you'd taken my money and grown it. Because what I really care about is whether you have a process that works, and I want you to feel pride about that. Now imagine you go to Vegas with some of my money, I stand over your shoulder, and you're perfect at counting the cards, figuring out the statistics and how they're changing, and you play every hand perfectly relative to the statistics — and you just get unlucky and lose some of my money. I'd say, 'Great job, take more of my money.' I will play that game all day long. The reason I take you and your listeners through that is: that's what innovation should be like, if you care about efficiency. Side note — innovation is really easy if you don't care about efficiency. You just find a lot of smart people, throw a lot of money at them, and you'll get some innovation. It's just not efficient. But if you want it to be efficient, you have to get people to abandon tying their pride to the outcome and tie their pride instead to a set of habits that tend to result in radical innovation efficiently. Now, easy to say, very hard to do. So you have to find hundreds of different ways to send them signals that you actually mean you want that, and reward them when they behave that way."

Ilana Golan: "Oh, this is so good. Oh my God, this is incredible. I love it — it's more about the habits and the formula, if you will, and when you can create that, it's incredible."

Astro Teller: "Yeah. If people want to hear more about some of this, our podcast is a deep dive into these stories — the creators, the people who've gone through X. I think it's really wonderful. But we also have a MasterClass certificate course, if people want to dig into the operations, some of these habits, and how to tie them together. I think the MasterClass on breakthrough innovation that we did would be a good way for your listeners to have that experience."

Ilana Golan: "Yeah, and MasterClass is incredible. And by the way, as I said, we're talking to the founder, David Rogier, today, so that's going to be super fun. But let me ask you one more thing, because you guys also had Google Brain, which had a substantial impact on AI as we know it today. And if I'm not mistaken, there was an article in the New York Times — I think around 2012 — that said something like 16,000 computers were needed to identify a cat, or something like that, right? So how did we get from there to here? Speaking of the formula — what was it?"

Astro Teller: "By the way, if people are interested, there's a deep dive on our podcast about this, but I'll tell you the teaser. This came to be because there was a professor at Stanford named Andrew Ng who had — in retrospect, it maybe seems like a really obvious discovery now, but at the time it broke artificial-intelligence orthodoxy. Neural networks were considered almost dead as a field back around 2009. There were a few researchers figuring out things, sort of like Andrew Ng was, but mostly it was considered a dead end for the field of artificial intelligence. But he noticed that every time he made his neural networks bigger, they got a little bit better. So he was like, 'Hey, I've got six points here. It looks like maybe we could just keep making them bigger and they'd keep getting better. I wonder how much bigger we could make them — and maybe that's just the solution to artificial intelligence.' The entire field of artificial intelligence didn't just say no; they got angry at him. So he came to us and said, 'Hey, I have this observation. What do you think?' And we said, 'Come on in.' That was an exploration that turned out to be right. He paired up with Jeff Dean, who came over from Google to be at Google X, and the two of them built up Google Brain. It started mainly around this observation: I wonder how much bigger we could make it. At the time, making it hundreds of thousands of times bigger than had ever happened in a university was not something a university could even run as an experiment. But at Google, we had the processing power and the expertise to actually try to make what, by today's standards, is a tiny neural network. By the standards of those days — late 2010, when we started Google Brain — it sounded crazy big. That's really how we got started: the bigger we made it, the more it seemed to work. And there were lots of questions, like how do you distribute the training of this thing across lots of machines? Because if we'd trained it on one machine, it would have taken forever — computers were a lot slower back then. So that's the little glimmer that started Google Brain."

Ilana Golan: "Incredible. I want to future-pace you a little bit: what do you think is coming? It's moving at a pace we can't even track — it's just mind-blowing. For those listening who are wondering what else is coming — and we also got a question like that on our YouTube channel — what do you think is coming? What's going to surface in the coming years that we might not even be thinking about yet? What do you see?"

Astro Teller: "I'll describe some of the things we're seeing that we're pretty confident are going to turn out to be a really big deal in the world. But before I do — another little soapbox, because it bothers me that there are so many visionaries who tell you how the world's going to be. Nobody. Nobody knows. That's the actual answer. You can explore into the world efficiently or inefficiently, but nobody just gets to tell you, 'Oh, I know how it is.' There's a whole class of people who think they know the answer, but if you average it all together, you get zero — *[laughter]* because if they all said the same thing, it wouldn't be interesting. To ground this: I think about the following all the time. I went to a museum exhibit in the late '90s at the Carnegie Museum of Art. They were doing an exhibit on the history of aluminum in product design and architecture. And in the corner of the exhibit, there was a hat stand, and it was made of wood. I was like, 'That's weird. Why did somebody leave a hat stand in this exhibit?' Then I went up, and there was a little plaque by it. It turned out that after an electrolysis process was discovered, aluminum suddenly went from being more expensive than gold per ounce to being a lot cheaper than gold. And one of the first things people made was a hat stand — and then they painted the hat stand to look like wood, because that's what a hat stand is. *[laughter]* The reason that sticks in my mind is that people hadn't yet imagined all the aerospace benefits of aluminum. The idea that we'd make beer cans and Coke cans out of aluminum, or make folding chairs — the lawn chairs from the '70s that many of us still remember — no one could conceive of those things. The thing I'm most confident about is that we don't know what the future is going to bring. Technology is going to surprise us, and the most profound impacts of the technology we can see today are things we can't even imagine — maybe don't even have words for. I guarantee you, if you went back to the ARPANET, the original beginning of the internet, and tried to describe to them this podcast we're having, *[laughter]* or all the other things — calling a Waymo through the internet — it would have completely broken their brains. That was not what the internet was for back when it was the ARPANET. But let me give you two examples of futures we're really excited about right now. One: right now there's about a trillion dollars a year of embodied value that just goes to landfill because we don't know how to get it back — not because it couldn't be recycled or reused, but because we don't know what we're looking at. Literally, if there's a potato chip bag going down a conveyor belt somewhere in the world, what's in that potato chip bag? You have no idea. If you have a camera, you can see it's a potato chip bag, but it's got 50 different chemicals in it. And if it has half a percent of the wrong thing, it will not only turn your whole recycling batch into goop, it might well ruin your recycling machine. So off to landfill it goes. A trillion dollars between plastic and e-waste — and that probably doesn't even include textiles or building materials, so it's probably a lot more than a trillion dollars. What if there were some magical way to look down on a conveyor belt as stuff was rushing by at 15 miles an hour and know every molecule in every one of the things zooming down the belt, and route it in just the right ways so you could get that embodied value back? Number one, someone's going to make a trillion dollars. But number two, you could get humanity back to being a lot more circular in our use of resources, instead of constantly burning dinosaur juice and digging up the world to get at the stuff that goes into the everyday objects in our lives. And so we've built that. It's called Matera. It's still somewhat early days, but it is very much working, and we're really excited about it. By the way, I managed to get through that whole thing without saying 'artificial intelligence.' Obviously, the way our system works, this thing looking down includes tools and techniques from machine learning — but it's so tiring when people go 'wiggle, wiggle, wiggle, AI.' AI is not the benefit. The benefit in this case is that you know what's in the thing *[clears throat]* so you can recycle it."

Ilana Golan: "True."

Astro Teller: "Let me tell you another story like that."

Ilana Golan: "I love that."

Astro Teller: "Right now, humanity makes six or seven trillion dollars a year in vats of various kinds. Most of it is chemical processes — think Dow and DuPont kind of stuff, oil refineries, those kinds of things. Humanity makes about 1% of that in fermenters — so, on the order of a few hundred billion dollars a year. And it could be so much more. Right now it's mostly alcohol, cheese, yogurt, some medicines, some cosmetics, but it's so expensive — especially if you're trying to get yeast or E. coli or algae, these tiny little self-replicating, carbon-negative machines that evolution has made for us, to make other things for us. In principle, they can make jet fuel for us. They could make things like building materials. There's not really an obvious limit to what we could ask biology to make for us. And we can now reprogram biology. This isn't X — the world has figured this out over the last two decades. So this is CRISPR finding a place on the DNA, and then Cas9 being able to snip into it and make a change on the DNA. But then what happens? You have a new yeast or bacteria or algae, and you hope that your reprogramming of its DNA is going to cause it to burp out something else you want instead of alcohol — maybe a medicine, let's say. Are you right? Who knows? Step two is: put it in a petri dish and wait. That's so slow that the entire field of strain engineering — trying to make these things better so humanity can move to biology being the main way we manufacture the raw materials in our lives — is completely bottlenecked around this issue of wait-and-see-what-happens. So X has spent the last eight and a half years building a simulator for small-cell biology and how it acts in its medium: How much food is there? What's the pH? What's the pressure? What's the temperature? And then, when you reprogram it, what is it actually going to do? How will it self-replicate? How much of the stuff you actually want will it burp out? And now this is working well enough — it's called A-Life — that we're offering it to manufacturers around the world who make things like MSG, or human milk sugar for baby formula, or cosmetics materials. We say, 'You give us your current yeast or E. coli or whatever, and we'll tell you a reprogramming that's wildly better. You don't have to pay us almost anything upfront — just give us 30% of the upside.' That's a pretty good deal, because if you don't succeed, cool. And then we're routinely making these things two to 10 times more efficient, which opens up whole new fields for them as those prices come down. Again, I got through that whole thing without saying 'artificial intelligence' — but obviously this process is a tight loop between wet labs for biology and artificial intelligence. So I've used these as examples. I hope your listeners can take away that the world feels topsy-turvy right now — but wouldn't it be incredible if some of the very things making the world feel topsy-turvy could also bring us these supercharging elements that make our lives radically better?"

Ilana Golan: "*[laughter]* Yes, absolutely. Yes to that, Astro. Let me ask you two questions that I think it'd be great to touch before we go. One: if somebody listening right now maybe lost their job, or is contemplating another job or another business, or is a little lost, scared of the future, trying to figure out what's next — the whole world of education and future work is changing in a massive way — what would be your tip to some of these people? Because they don't have the backup of a Google, with millions to spend on innovation, but they do want to create something big, or pursue a dream, or test things out and start experimenting. What would you say to them?"

Astro Teller: "I'd start with: it's important that we have compassion for everyone being on their own journey. There are people who look like they've got it made who actually have big challenges, and people who look like they're really struggling — or in whose circumstances we would really struggle — who are actually footloose and fancy-free. So I just want to recognize that it's really easy to armchair-quarterback somebody else's life. That said, I see most of the people even here at X, and most of the other people I meet in the world, operating at five or 10% of their personal capacity. They're weighed down by — mostly, I think — some version of fear. That doesn't mean none of us have anything to fear. But letting fear control you and your choices is so destructive to our ability to get the things we want in our lives. Which brings us back to this issue I was describing: being able to manage your psychology, to get into a neutral position from which you can be wise and make choices that might be complicated, might have some risk in them, but where you're not letting your fears, your sense of guilt, and other things grab your steering wheel and whip you around the road. That's what I see most people spending most of their lives doing. And what I just described — open-heart surgery is easier to do than actually building a deep understanding of yourself and being able to manage your own psychology in a really great way. So I'm not saying it's easy, but that's the work. For anyone I'm coaching, it's all about that, because once you can manage your own psychology, everything else is kind of easy by comparison. And until you do, it doesn't matter whether you've got Alphabet's backing or not — it's pretty easy to screw things up."

Ilana Golan: "That's so good. Do you think there's something in your past, Astro — something some people don't know — that built you into who you are today?"

Astro Teller: "Probably most of your listeners don't know anything about me. I'll give you an example — you could go digging if you want more. I grew up in a family particularly focused on intellectual prowess. My father's father was Edward Teller, the father of the hydrogen bomb, who helped build the Manhattan Project, started the nuclear submarine process for the United States with Admiral Rickover, and actually started the Star Wars Initiative, the space-based defense system. And my mother's father won a Nobel Prize in economics in 1983."

Ilana Golan: "Oh my God, that's a scary family to be in. *[laughter]*"

Astro Teller: "When I was a kid, the yardstick was how smart you were. And I love him dearly, but my brother is much smarter than me. I had lost to my younger brother by the time I was probably seven and he was five — it was super clear I was the dumb one in the family. That was really complicated for me. At the time it felt like an entirely bad thing, but I think I got really lucky in a funny way, because I learned at a really young age that how smart you were wasn't how I was going to win. So from a very young age, I got interested in what else I could do. I remember being really attached, around seven or eight — do you remember those Avis ads way back in the day? 'We try harder.' *[laughter]* There was a sense — it turns out effort is not a good way to try to win, so I'm not advocating for it — but when I was a kid, there was a period where I thought, 'We try harder, I'm going to be like Avis.' Again: don't do that. Work smarter, don't work harder. But I've learned a lot, because I felt from an early age that I had to go figure out, on my own, new yardsticks to define my own self-worth — to figure out what I was really good at. I think sometimes we take too much of what society says you're good at at face value, and I got to skip that part of the process. So I think that really helped me, weirdly."

Ilana Golan: "Huh. Do you know what built this level of confidence — the confidence that now lets you build something like Google X, and these kinds of innovations, and continue to build yourself, and know 'I can do this'? How did that become so strong?"

Astro Teller: "I'm not completely sure, but let me give you another story to see if it helps. I was in college, still probably at the tail end of my 'we try harder' phase, and a doctor sat down with me. I thought I was having an ulcer at the time — this was back when people thought ulcers were caused by stress. (They're not, for anyone who still thinks that's true.) And the doctor said — in retrospect, this was kind of irresponsible of him — 'You're going to die by the time you're 50 if you go on like this,' because I was pushing myself too hard. Despite it being an irresponsible thing to say, I really heard him. I had a couple of weeks where I thought really hard about it, and I made a new deal with myself that I think has stood me in really good stead ever since: I am not going to hold myself to outcomes. I can't hold myself to the outcome of being set up where my grandfathers were — that's a few hundred people in a century who ring the bell like that. I can't hold myself to that standard; it was literally driving me crazy. But I have to hold myself to some standard. So I said, 'I'm going to hold myself not to the outcome standard, but to the process standard. If I am a ferocious learner, that's what I'm going to be proud of. If at the end of every day I've been open to feedback — not just to hearing it, but to taking it and trying to get better at anything — if I stay really curious, then I'm going to trust that, with some luck involved, the outcomes will be what they are. I bet they're actually going to be pretty good. I just don't care about the outcomes anymore. I only care about whether I personally showed up today as a ferocious learner — as a really open, generous, kind-hearted human who is in receive mode and improve mode. If I do that, I win.' That's just my new definition of winning. And I think — I don't know, I got lucky — but I think there's a lot of truth in living your life that way, because I have complete control over it. I don't always do it perfectly, but at least I have control over it. I don't have anyone to blame but myself when I don't show up that way. There's no luck involved in that. I think that's really helped me to be confident and happy."

Ilana Golan: "That's incredible. I don't know if you notice how equivalent this is to the process of innovation you're building. It's pretty incredible to hear, because it's those imperfect steps — 'I'm going to take them, I'm going to look only at the formula' — and sure enough, at the end of the day, ta-da, you're running the whole moonshot, CEO of Google X. That's incredible to hear, Astro. This is such a fun story. Thank you for sharing it."

Astro Teller: "This is so cool. I think it's worth saying — both for people who are running something and for people who are part of an organization. I appreciate the kind shout-out, and I obviously do my part here, but this is not a top-down organization. I am a supporter and a nurturer of the awesome people who work here. As you were talking about earlier, they have to do the hard yards at least as much as I do. I often end up being the one on the podcast who talks about it, but this is not the Astro Show, with everyone just implementing my vision. I feel pretty strongly that that's actually the right way to do innovation. So I'm not rejecting your compliment, but I think it's important that your listeners appreciate it's actually about all of them — which, again, is why we did the podcast. The podcast is not me saying, 'Let me astrosplain to you how X works.' It's me, as the host, asking questions and trying to pull out of these awesome human beings who've gone through the Moonshot Factory: What was your journey like? What were the hard parts? Where did the breakthroughs come from? Now that you're on the other side and you're winning, what does it feel like? Or, if you killed your project, how did that feel — what happened to you afterward? I think those people and their journeys are actually more important than what the leader is doing."

Ilana Golan, host of the Leap Academy Podcast: "Oh, this is powerful, and I'll end with it. It's worth listening to both — the podcast and the MasterClass are very, very different. And in Leap Academy, we always say it's not about what you make, but what you make possible. I think it's really fascinating to see what you, Astro, as a human — but also Google X — are making possible in the world. It's just really fascinating. Thank you for sharing this, Astro. This was fascinating. I could probably talk to you for hours."

Astro Teller: "It was my pleasure. The time went super fast, and I really enjoyed it. Thank you so much."