Pod Insights Decoder Recode Media: Inside the AI Gold Rush About this episode Today – we’ve got a treat for you. We’re going to run a special episode from our friends over at Vox. Peter Kafka and his team just wrapped up a special 3-part series on AI.
AI has captured the imagination of Silicon Valley. In fact, in the last few months, I’ve talked to both Google CEO Sundar Pichai and Microsoft CEO Satya Nadella about AI after they announced new AI-powered search products. And in the middle of the frenzy, it's hard to tell what's really going on. What exactly is AI, how does tech plan to re-design the world with it, and why are a bunch of smart people very, very worried?
In this episode, they’re diving into the gold rush around AI. Figuring out what’s just hype, meeting the VCs that are hungry to invest, and finding out if there will be room for startups, or if the giants will just own it all.
If you’re a Decoder listener, this is right up your alley. Thanks to Peter Kafka and Vox.
Learn more about your ad choices. Visit podcastchoices.com/adchoices
Transcript 9,502 words · 11 speakers
0:00 Speaker 00 Give you and your team more time to focus on the work that matters. This is AMEX Corporate, one intelligent platform for company spending. With smart expense management to help capture details, categorize spend and streamline expense submission and approval, there's less manual work for your team and greater visibility and control for you. Built to simplify, designed to scale. Get started at go.amex slash AMEX corporate. Terms apply.
0:30 Speaker 01 Flowing ad budget on metrics that look great till the CFO sees them. That's bull spend and marketers are calling it out in dashboard confessions.
0:40 Speaker 02 I remember telling my boss, it'll be good for the brand. When leads were slow. Yeah, it, it wasn't.
0:47 Speaker 01 Cut the bull spend. LinkedIn lets you target by company, job title and more. Advertise on LinkedIn. Spend $250 on your first campaign and get a $250 credit. Go to LinkedIn.com slash campaign. Terms and conditions apply.
1:02 Speaker 03 Hello and welcome to Decoder. I'm Neelai Patel, editor in chief of The Verge. And Decoder is my show about big ideas and other problems. We have a fun one for you today. We're going to run a special episode from our friends over at Vox. Peter Kafka and his team just wrapped up a special three-part podcast series on AI. Obviously, AI has captured the imagination of Silicon Valley. In fact, in the past few months, I've talked to both Google CEO Sundar Pichai and Microsoft CEO Satya Nadella after they both announced new AI-powered search products. In the middle of all the frenzy, it's hard to tell what's really going on. What exactly AI is, how tech plans to redesign the world with it? And why are so many smart people so very, very worried? In this episode, Peter dives into the gold rush around AI,
1:44 Speaker 03 figures out what's just hype, meets the VCs that are hungry to invest and finds out if there's going to be room for startups or if the giants will just own it all. If you're a Decoder listener, this episode is right up your alley. Thanks to Peter Kafka and the Vox team. Here we go.
2:03 Speaker 04 I am an adult, so sometimes I use LinkedIn. And when I'm on LinkedIn, I see the same stuff I think you do. There's job announcements, people talking about promotions. You get a few likes, a few congratulations. That's it. Well, that is not what happened to Renata Nyborg when she posted something on LinkedIn a few months ago. Nyborg was starting a new gig. She was thinking about starting a company. She wasn't sure what it would do, didn't even have a name yet, but she wrote about it on LinkedIn anyway.
2:31 Speaker 05 I said, I'm joining AI Fund as a founder in residence. I'm going to be working with Andrew Ng, who's been a long-term hero of mine to work on something in the health space.
2:41 Speaker 04 Some quick translation here. Andrew Ng is a big name in artificial intelligence. He runs an investment company called AI Fund. And so Nyborg was saying basically, hey, I'm eventually going to do something. It's going to be something involving health and something involving AI. So yeah, she got the likes and congratulations, but she got a lot more because it turns out that she was essentially putting out a bat signal for people with money who wanted to put that money into tech's hottest new thing.
3:09 Speaker 05 It was really funny. My inbox went bananas. Even though no one knew what I was working on and I hadn't decided what I was working on yet, I received inbound interest from dozens and dozens of VCs, some of whom were extremely persistent. I was even at dinner with a friend last night who said, yeah, I was just getting so many VCs calling me saying, hey, you know, Renata, can you please try and get a coffee set up with me?
3:34 Speaker 04 People who want to give you money are reaching out to your friends so they can get in touch with you. They're bothering your friends because they're so desperate to give you money, even though they don't know what you've built and you don't know what you're building.
3:48 Speaker 05 Correct.
3:49 Speaker 04 Did you expect that?
3:51 Speaker 05 No.
3:52 Speaker 04 I just want to spell out how unusual this is. Sometimes at the height of a Silicon Valley hype cycle, you hear about people literally investing based on a napkin sketch, just the bare bones of an idea. And this is pre napkin sketch. There is no idea.
4:10 Speaker 04 To be fair, Renata Nyborg is not your average startup founder. She was making apps for mobile phones long before the iPhone came along. She was an executive at Apple. Her last job title was CEO of Tinder. So she's a good bet for VCs. But at this point, Nyborg wasn't even looking to raise money.
4:27 Speaker 05 No, that wasn't the purpose at all. I was very surprised and initially I really wanted to focus on finding my problem and prototyping. And so I didn't respond to most of these people.
4:38 Speaker 04 Instead, Nyborg figured out what she actually wanted to work on and she decided she wanted to solve loneliness.
4:46 Speaker 05 I think it is the biggest threat to our health, to democracy and a bunch of other things. But luckily, there is a medicine, which is human connection. So I'm working on that.
4:55 Speaker 04 And she named her company Amory, as in amor. That's love plus AI. And the first product she's working on launching is a subscription-based AI relationship coach.
5:07 Speaker 05 I've been trying to explain to people more is the movie Ratatouille. The movie Ratatouille is about a rat who knows everything there is to know about food. And he helps a young man who wants to be a Michelin star chef cook fantastic meals.
5:18 Speaker 04 You're gonna sit on top of my head and show me how to chop stuff.
5:21 Speaker 05 And so you're a perfect chef.
5:23 Speaker 04 And this will apply to health and loneliness.
5:26 Speaker 05 It's a coach that is basically helping you find solutions to practical problems that you have. So this could range from I'm struggling to get over my breakup, can you help? Very practical things like I keep seeing iPhone memory, photo memories coming up on my phone. What do I do about it?
5:44 Speaker 04 Nyborg was cagey when I asked her how her customers are gonna actually interact with this relationship coach. And she's still working on it. But she did say you'd be able to text with it.
5:56 Speaker 04 And this brings up a question. Why does this product need to be an AI texting thing at all? Why couldn't you just talk to a human? Because those things exist. And one answer, obviously, is that AI is supposed to be cheaper. But Nyborg also says that according to her own market research, her customers, or at least some of them, don't want to talk to a human. They want to talk to an AI.
6:19 Speaker 05 Young men actually felt much more comfortable interacting with some of the futures of this product because they knew it was AI. And it allowed them actually to explore some of these vulnerable emotions for the very first
6:31 Speaker 04 Dude dudes dudes are more likely to talk to a person they know is not a person because they feel more comfortable knowing it's a computer.
6:41 Speaker 05 As a starting point, it's not supposed to give you a fake friend that you talk to for the rest of your life. It's very much focused on giving you advice and ideas for how to do things in the real world.
6:52 Speaker 04 So now our company had a name and a product and it was time to actually start raising money. When you decided you were going to build, how long did it take you to to raise that money and how much did you raise?
7:04 Speaker 05 I'm not disclosing how much we raised, but I raised the round in 24 hours last week.
7:10 Speaker 04 Okay, so put that in context for people who are not hanging out in startup land. How common is it to raise money to start a new company in 24 hours?
7:20 Speaker 05 Pretty uncommon, especially at the moment.
7:23 Speaker 04 What did that tell you about about the world in January of 2023?
7:28 Speaker 05 I think it told me that people are really interested to get a foot in the door with AI. This is a time where people that maybe put money into crypto or the quote unquote metaverse and so on are looking for the next thing to double down on.
7:41 Speaker 04 And this is what doubling down looks like in Silicon Valley. Investors are supposed to pour $43 billion into AI this year. That number is supposed to get to 98 billion by 2026. And one big question is whether all this money is going to pay off for individual startups like Renata Nyborg's, or whether the real winners in AI are going to be a handful of big companies that are already established. Today, I'm going to talk to two people who are trying to figure this out. First up is Druar Berman. He's a venture capitalist at Innovation Endeavors. It's a company he co-founded with Eric Schmidt, the former CEO of Google, and he's one of the guys throwing a ton of money at AI companies. And the reason he thinks he'll get it right is because he's done it before.
8:24 Speaker 04 Some people who are psyched about AI say it's going to be a new platform like the App Store and on the iPhone, which led to the creation of some big new companies like Uber. And Druar Berman invested in Uber. So maybe he's going to find the AI version.
8:46 Speaker 04 So the positive version of the AI moment we're in is the one you believe in is this is another big important shift like the iPhone and mobile was. So if that works out for you personally, that's great. And for the world supposedly that would be good. The cynical version is this is the blockchain. This is web three. This is a thing people were very excited about a couple of years, just a year ago, you have people having really serious conversations about web three and how it's going to change the world. And, and now there's NFT conferences that no one is attending. There's just empty hallways. Tell us why you believe this is the mobile version of the future and not the blockchain version of the future.
9:26 Speaker 06 I think it's a much more exciting point in time than the mobile version. I mean, the mobile version was an interesting time because it provided sort of a new form factor that allow you to carry a computer with you and do certain things. I think we are now standing in a completely different time where we, we are now, we've now been introduced to a foundational intelligence block that has become available to us is basically can lean on all the publicly available knowledge that humanity has extracted and documented. And it allows us now to essentially retrieve all this information in a way that wasn't possible in the past. It's flexible. It can adapt. It can learn and largely it performs way better than anybody expected. Now, two question about web three and, you know, I've
10:11 Speaker 06 been doing this for quite a while now. So I've seen all the different hype loops and so on. I think every once in a while you do get, you know, selling hype cycles that are based on completely wrong things. I think in my opinion, and I might be not the most informed person about blockchain. I'm sure a lot of people would disagree. I actually looked quite deeply in the early days of blockchain to try and understand what was there. And I couldn't find. Really interesting application that somebody could use. So this is not one of those cases. We've already been developing it and seeing the applications in the real world. And I think we're just getting started.
10:45 Speaker 04 You seem smart. You've got a track record. You're working with the former CEO of Google. You seem like you know your stuff. You've got a good argument for why this stuff is important. Again, a year ago, I was talking with other smart people who had track records and seemed very convincing. And they were telling me blockchain or web three or some version of that was the future. What's a useful way for a person who's not deep into this to go, all right, that was BS, but this is the real thing?
11:11 Speaker 06 Yeah, it's actually fairly easy. I mean, go to chat GPT and have a discussion. I was doing that last night. I was thinking about this interview and preparing and I was, my, my nine year old daughter came down. She couldn't fall asleep. So she came down and said, okay, let's, let me show you something. And, you know, we started playing with it and the amount of things that we could do was just, you know, it was just almost like magic. I asked chat GPT to write a story for her and then I asked her to adjust the story to be in the style of Dr. Seuss. And I asked it to adjust the story to be in the style of South Park and sort of, and then I provided more and more constraints and it kept evolving and changing.
11:51 Speaker 06 And so, you know, we are seeing the same thing now across many different domains that are just starting to discover what's possible now. We're seeing the same in healthcare. We're seeing the same in education. We can now personalize and summarize very complicated concepts to make them available and accessible to anyone. And it's available for everyone. You can just sign up and, and see what it is.
12:10 Speaker 04 So there's stuff you can use right now. And I guess to me, that's the biggest, I mean, obviously there's lots of other distinctions, but when we were talking about web three, you'd show someone a picture of a monkey smoking and saying, this is worth a lot of money and we're going to make movies out of it. And I think neither of those things turned out to be true, but even if that was the case, it was sort of hard to understand what we were doing beyond that. So let me, let me talk to you about stuff you are excited about. You have a lot of money to deploy into AI and you've been doing it. What are specific ideas and use cases that you are excited about that you think this is money I can deploy right now because this is a real thing?
12:48 Speaker 06 Yeah, I'm excited about anything where initially we could use this kind of technology as sort of a co-pilot or a partner for an expert, right? And I think we're gonna start seeing that across many different domains, anything from drug discovery and use that to design better medicine, right? I think that's one of the, one of those use cases is exciting for me. I'm also excited about new type of companies that build on this sort of foundational layer of what we call large language models with more specialized data. So you can think about a new language models that is largely targeting specific domains like healthcare and, you know, is trained on very specific data that maybe is not accessible to OpenAI and can allow us to answer certain healthcare questions, to diagnose different
13:33 Speaker 06 diseases, you know, faster, better and cheaper and so on and so forth. And then, you know, we've also invested in quite a few companies. We have companies that already use generative AI to generate better text and copy and, and stories. So we have companies that use generative AI to look at all the information that's available in the financial world and help people consume this kind of information.
13:53 Speaker 04 People are very excited about this, um, and you believe this is real and not BS, but there is a lot of stuff coming at you. What are examples of stuff that people are bringing you that you don't want any part of because you don't think it's real? You think this is made up or, or can't deliver or isn't that interesting? What, what stuff do you not want to touch?
14:13 Speaker 06 Yeah, I, I would say the sort of the obvious stuff or things that do negative things in the world that I would not want to touch.
14:19 Speaker 04 So you don't want an automated bomb-making machine?
14:21 Speaker 06 Yes, and I think we
14:22 Speaker 04 Fair enough.
14:23 Speaker 06 We, we do not want that. But I think in general, my approach in, to investing over the years has been one that is trying to look at very specific constraint type of problems. I've been worried about people who are trying to create the next big thing in something where it's clear that we are not there yet. You know, autonomous vehicles, which we sort of, we saw the wave of autonomous vehicles roughly. I don't know, seven, eight years ago, where there's a lot of companies that promised us we'd have robotics running around every corner. That largely didn't work. What did work was a much more constrained version of it. This was a company we invested in called Gadic that basically decided to build a very specific applications that they looked at the problem of how do we move goods from
15:06 Speaker 06 one distribution center to another, where you could actually sort of train on just this specific route. And they are driving autonomously now for Walmart and a few other big customers.
15:16 Speaker 04 It seems like there's an interesting tension there because you say, I'm excited about this stuff. This is going to change the world. I'm looking for companies and leaders who can change the world. You also say, I'm looking for people who can work on a constrained problem and not trying to boil the ocean and don't tell us we're going to have self-driving cores on every corner, but we could use them in this very limited way. How do you balance those ideas out?
15:41 Speaker 06 I think you should, as a leader for all of those companies, I think you should do both. I think a lot of people perhaps overestimate what they can do in 18 months, but underestimate what they can do in five years. I want them essentially to understand what's a five or 10 year plan of how this could actually change the industry. But I also like to see a very concrete plan on how we're going to go into the next step and achieve something meaningful, because a lot of those companies, they need to raise money, they need to show their employees that there is something real behind this vision. And actually, that's hard to do if you just build something for 10 years that nobody can actually see. So I think for the best leaders basically try to balance those two things and build a real company around them.
16:18 Speaker 04 One thing I've heard a lot of people point to saying this works right now, it has real meaningful consequences, is using generative AI to write software. Does that make sense to you? Are you excited about that idea?
16:32 Speaker 06 Yeah, there are only solutions out there. If you look at GitHub Copilot and others, that could look now at all the pieces of code that were available to them to essentially learn and then reason based on the context, what you've written before and so on, to essentially allow you to complete your code. So the GitHub Copilot, for example, is already out there, it's generating great revenue. This is already saving a huge amount of time for developers and largely leading them to write better code. But what's interesting is also those systems are learning and getting better. So looking at every type of recommendation you decide to take and every type of recommendation you decide not to take, what should make the system better? So those systems become better as more people use it and as they use it more.
17:12 Speaker 04 It's really interesting to think about AI being trained to write code because for most of my adult life, I've been told that a good job for people to have is to figure out how to learn how to code, right? And now it's almost a punchline. And now you've got computers that can do that. And this was supposed to be lucrative, sustainable, white collar work. Or now we're saying, computers can do this work. Should people be worried about about AI replacing them, whether it's coding or other work?
17:44 Speaker 06 I think society will need to adapt. I think we're sort of just starting an era where a lot of those systems are very, very powerful and allowing us to do things that we never thought could be possible. By the way, largely, we don't yet understand what is fully possible. We don't also fully understand how some of those systems work. So I think the answer is, yes, you should not necessarily be worried, but I think you should think about how do we adapt to this new era where we can do new type of things.
18:12 Speaker 04 I want to get a little point here, though, right? Eventually, these companies you're investing in, whether you're investing in or not, that are writing code, for instance. The pitch to someone who's going to buy them, use them is, maybe it's, we're going to help your existing workforce produce more work and produce better work. But inevitably, part of that pitch is, you're going to have to spend less on coders because we've got this machine that can do it for you, right? This is the John Henry parable. Isn't part of the promise of this tech inevitably that it does cut costs by reducing the number of people you need to make something?
18:51 Speaker 06 I don't view it that way, although you could look at it that way if you want to. I think if you look at even the GitHub Copilot project, it basically allows us, for now at least, to achieve significantly more than we were able to achieve with the same amount of developers. So you can think about writing faster, better code in a way that wasn't possible up until even six months ago. In the beginning, we're going to see more of those copilot type of software, right? So we're going to see more AI as a partner to a lot of those.
19:20 Speaker 04 AI is assisting you, right? I hear this. I hear this idea a lot. And it sounds like people believe that, but it also seems impossible to believe that people aren't saying, sure, they could be the co-pilot, but why don't we just make them the pilot? Why should humans do this work if computers and software and robots can do it? We shouldn't have them doing that. We should replace them. It makes much more sense.
19:43 Speaker 06 Yeah. And I think we've already been seeing some of this tension over the last even 10 years in using AI. I think largely in a lot of those domains that we've been very used to having those being run by experts, including biology and material science and so on. We've started to see more reliance on, on data and AI to essentially guide us through this process and do things significantly faster. And I think this kind of technology allows us to power boost this cycle even more. And it will get better as we, in some of those domains, as we take humans out of the loop. I think that would mean that in a lot of those domains, humanity will need to adjust. I think it would allow us to do greater things. It would allow us to dream bigger.
20:24 Speaker 06 And the exciting part of it, which I see is that it would allow us to solve a lot of the imminent problems that we need to solve as humanity. I mean, we've all seen how important it is to come up with a solution for COVID in a very short amount of time. I live in California and I see the fires spreading here every summer now. And, you know, this is, you know, this is just one other example of climate change, but we know that we need to do significantly better. And if you look at the, historically, the pace of innovation we've had as humanity, and you just sort of extrapolate from it as to where we, we would be in the next 10, 20 years, it's not going to be enough. So this kind of system should allow us to move significantly faster. And for me, that's exciting.
21:02 Speaker 04 If I summarize what you're saying is, yes, some people are going to lose jobs, but we're going to move faster, we're going to make new stuff, and we're willing to live with that. That will be one of the downsides is that some people will not have jobs, but the net result is better and positive. And we're, that's what the future we want. Is that a fair summary?
21:22 Speaker 06 Yeah, and I think some people will lose jobs, some people will adjust and get new jobs. So we have a company called Canvas Robotics that is developing a new type of robot for the construction industry. And they're actually working with the union to train the workforce to essentially use this kind of robot. And I mean, guess what? A lot of those jobs that a lot of those technologies replace are not necessarily the jobs that a lot of people want to do anyway. So I think that we're gonna see a lot of new capabilities that allow us to train people to do much more exciting jobs as well.
21:52 Speaker 04 I'm talking to several people who are real skeptics of AI. But rather than repeat their arguments, I'm curious what you're skeptical about when it comes to people making claims for AI or concerns people have AI. Which ones are resonant for you?
22:06 Speaker 06 I mean, I think we have still a lot of limitations. I think from a capability perspective, there are still domains where there is not a lot of publicly available data, so we don't necessarily know how the performance of those models will work. I think there are some limitations around what I see as transparency. Again, we don't understand why those models work the way they are. Like, what makes OpenAI or GPT write a South Park-like story? And I can ask it to make it more funny, and it will. And, you know, we don't necessarily understand it fully. There is a lot of biases concerns, because again, we are leaning on a lot of existing data. So, you know, if the existing data is biased in some ways, this might generate bias in the future. And then there are some ethical concerns, right? Like, you know, regulators will need to address, you know, how do we prevent people from doing
22:51 Speaker 06 really bad things using this kind of technologies?
22:53 Speaker 04 I'm sure you've seen this. There was a letter published recently by a lot of serious people saying, hey, hey, we gotta slow down this AI stuff. We just don't know where it's going. Let's take a six-month pause. I don't think anyone's gonna listen to them. But do you do you take that that line of argument seriously that we're plunging headlong into something and we don't fully understand what we're doing and we should slow down? We shouldn't try to speed up?
23:17 Speaker 06 Well, I think that even if we wanted to do it or we thought it was a good idea, I think it's not feasible at this point in time. I do think that regulators will need to address the concerns here, right? And I'm not a regulator, so it's not up for me to to decide how this would happen. But I think regulators will need to address it. I think for as far as I can. You know, influence this sort of making sure that things are going in the right direction. You know, we think about seriously with the companies that we are working with, right? So essentially, how do we, how do we get a, how do we make sure that there is no negative externalities to the products we're putting out there?
23:53 Speaker 04 I'm gonna stop you right there. What does negative externality mean?
23:57 Speaker 06 Avoiding bad things from happening unintentionally. So yeah, I think for, for us as investors, as people who are actively involved with the companies that we support, we need to provide a governance to make sure that the products we put out there are doing positive things in the world and avoiding negative unintentional consequences as well.
24:16 Speaker 04 Your bio says you invest in ideas that are, this is a quote, "driving exponential rates of change in their industries." So that's pretty broad. Which, what industries do you think are going to see the most exponential change from AI?
24:28 Speaker 06 So the, the things that we, we've been focused on is a lot of industries that are encoded in the physical and biological worlds. And I think those historically have benefited less from technology and are now starting to benefit significantly more because of what we talked about earlier, because a lot of those industries have been largely driven by humans, by scientists going through very slow cycles and sort of building the next step. And what's exciting is that a lot of them have been digitized or increasingly getting more digitized with increasing amount of sensing that is available to us. We now have the compute power to essentially take all these massive amounts of data, such as what we just talked about, and infer from it and come up with sort of new ideas based on the
25:10 Speaker 06 data. And then we can engineer things back on the physical and biological world in a way that wasn't possible even a few years ago. So we can use robotics and automation to build things faster. And we can use synthetic biology to manipulate living things. And I'm also excited about the foundational layer. So, you know, the next companies that basically allow us to build the next set of tools, the open eyes of the world, you know, those kind of companies that are generating the next set of technologies to allow us to keep changing the world.
25:40 Speaker 04 Up next, inside the brains of Silicon Valley and who the Valley thinks is really gonna win the AI moment.
25:55 Speaker 01 Flowing ad budget on metrics that look great till the CFO sees them? That's bull spend. And marketers are calling it out in Dashboard Confessions.
26:04 Speaker 02 I remember telling my boss, "It'll be good for the brand," when leads were slow. Yeah, it wasn't.
26:12 Speaker 01 Cut the bull spend. LinkedIn lets you target by company, job title, and more. Advertise on LinkedIn. Spend $250 on your first campaign and get a $250 credit. Go to LinkedIn.com/campaign. Terms and conditions apply.
26:26 Speaker 07 This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required. Compatibility and availability varies 18+.
26:51 Speaker 01 Are your ad campaigns lighting up the dashboard? But not the pipeline? That's bull spend. And marketers are calling it out in Dashboard Confessions.
27:00 Speaker 02 My boss asked for results, so I opened my dashboard for the only positive-sounding metric I had: impressions.
27:08 Speaker 01 Cut the bull spend. See revenue, not just reach. LinkedIn delivers the highest return on ad spend of major ad networks. Advertise on LinkedIn. Spend $250 on your first campaign and get a $250 credit. Go to LinkedIn.com/campaign. Terms and conditions apply.
27:22 Speaker 07 This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required. Compatibility and availability varies 18+.
27:52 Speaker 04 Jessica Lessin is the editor in chief of the technology news site, The Information. Before that, she covered tech for the Wall Street Journal. She knows tech inside and out. She's got her finger on the pulse of the VC class in Silicon Valley. And right now that pulse is racing. And I was hoping she could explain why. Jess, welcome.
28:13 Speaker 08 Thanks for having me, Peter.
28:14 Speaker 04 Thanks for coming in. I wanted you to translate AI for me and AI and Silicon Valley and sort of the tech mindset. Obviously, your reporters on staff are covering the AI moment, the AI boom. But you're also just in the mix. You're with founders, with entrepreneurs. You've got a decent sense of what they're thinking about, sometimes what they might say when they're not being recorded, when they're not on the record. Let me just start big picture. How would you describe Silicon Valley's attitude towards AI right now, April 2023?
28:44 Speaker 08 Excited is actually, in a corny way, almost the word that comes to mind. I think there is a real energy to both the activity at many startups and also in the big companies, you know, we've got Microsoft fighting Google, tripping over each other to get their press release out, you know, with the latest version. And there is a real energy and excitement. There's, I'd say also, you've really felt the concern growing a little bit, like, what are we unleashing here? But overwhelmingly, I'd say to start, you know, it's felt like this kind of breath of fresh air.
29:24 Speaker 08 There is something exciting to talk about that isn't crypto.
29:31 Speaker 04 Put this in context. Have you seen other Silicon Valley hype cycles before? People get excited. It's the nature of Silicon Valley, I think, both because it's supposed to be forward looking and optimistic, and also they need a thing to be excited about professionally. Does this feel different than other booms or boomlets or would be booms you've seen in the past?
29:52 Speaker 08 It does. I mean, there are some similarities. So maybe to start with those, you know, every day I hear about a new venture fund that is pivoting to AI or, you know, dressing its latest fund exclusively in AI. And that, to me, feels very crypto or even very VR, AR, one of these other trends. Now, there are very big questions about the business models that I think actually have a lot of investors pretty nervous as well. And so it feels like we might be transitioning from the phase one optimism to the phase two, a bit more cautious. But I would say it does feel palpably different from crypto in that, you know, it's not something wonky that no one necessarily understands,
30:34 Speaker 08 although not that we all understand it, but so many people are using these tools. And so there's a familiarity with the use case that I think is helping drive it.
30:44 Speaker 04 It seems to me that maybe it's a coincidence and maybe it's not that the crypto boom died out. If we were talking a year ago, we would have been very interested in crypto and web three. But I think by this summer, that was pretty much exhausted and there were lots of different markers for it. And almost as if on cue, within a month or so, it seemed like and throughout the fall, everyone was getting progressively more excited about AI. Is that total coincidence? Or is there a link between the end of one bull phase and the beginning of another one?
31:17 Speaker 08 There's a link. I mean, this is a place of cycles. And I think that can lead to a lot of hype, but it also leads to a lot of investment and a lot of innovation. So there is no question that the world out here has been looking and is always looking for a new platform, a new set of technologies. And as crypto was falling, that absolutely fueled AI in a lot of ways. But but but it is also more than that. One of the reasons AI is so exciting right now is there've been these sort of sleeping giants that have been pushing the field forward. And OpenAI, you know, a company run by Sam Altman with a strong alliance with Microsoft, I think, is very important here because I credit
32:01 Speaker 08 A lot of the enthusiasm around this cycle with sort of their business decisions to start releasing product, to market it to developers, to try and build a very broad bottoms up kind of network of partners, I think in large part to counteract the bigger platforms and to kind of play a game they can't. And so I think when we look back on the timing of this part of the cycle, their role in it will be very important.
32:30 Speaker 04 I was going to ask you about about Google, Microsoft, OpenAI, but but something you just said is really intriguing, something I've been thinking about a lot. What did OpenAI do? What can they do that they can do sort of as a startup that a Meta or a Google either can't or thinks it can't?
32:47 Speaker 08 It's a wonderful question, and I think a lot of that is unknown, but is in the eyes of the decision makers at these companies who are trying to, you know, avoid unwanted regulatory scrutiny. You know, Google is one of the most closely watched companies on the planet from regulators around the world concerned about its growing power. And for better or worse, and it may turn out worse, you know, Sundar and the leadership there has decided to be very, very cautious. And, you know, our reporting has shown had years of projects under the hood that just didn't progress, or that that came out in kind of compromised ways because of that. And even if you use their chat tool today, Bard, it comes with a very big disclaimer on
33:31 Speaker 08 top around, you know, accuracy, trustworthiness, sort of saying, we want you to play with this, but don't let it backfire.
33:38 Speaker 04 They're so big, they can't afford to move fast and break things anymore.
33:41 Speaker 08 Yes. And and look, it's a very uncomfortable landscape with these chatbots kind of synthesizing some real information, some made up information, acting so human like in their conversations, we can forget that. I mean, there are reasons to be cautious. OpenAI has not been. Now, now, again, I think they have guardrails as well. It's not that they think it's the Wild West, but it with so many things from Dolly on the images all the way up through chat, you know, they've really.
34:12 Speaker 08 Kind of taken, and this seems very Sam Altman, right? Who came from Y Combinator and the startup world out here. Very, let's win over the hearts and minds of developers and get our hooks into as many products as possible.
34:25 Speaker 04 It from afar, it feels like it has that old Silicon Valley ethos. We're in the garage, we're playing around, we're going to put stuff out. It's software, so inherently we know that it will have bugs and it will get better over time and, and our users expect that. And so I, I see the appeal of that to a lot of people who are in the valley. On the other hand, this is a different kind of tech, right? And, and the world's different than it was when Silicon Valley got it started, even when these internet entrepreneurs were starting. It's really interesting to me that Microsoft, which is not a startup and is, is dodgy and entrenched in the existing software world as you can be, has taken a bet on this and has put OpenAI product out in the world and said, pretty much the same thing as Google, like, this could all be wrong, let's see.
35:10 Speaker 04 How, how does, could you give the insight into Satya Nadella, who's the CEO of Microsoft? It seems uncharacteristically risky for any company this size, let alone Microsoft, to be playing with this stuff it doesn't fully have control over yet.
35:24 Speaker 08 It's an excellent question. And again, I think our reporting shows he felt he had no choice because they were behind. And Microsoft Research was not making the kind of progress Satya wanted to see or that he felt, you know, others were. And along came this small company that needed servers and a lot of them. And in a partnership forged over cloud computing, you know, OpenAI got the infrastructure and the capital it needs. And by the way, it needs a shit ton more, so we have to figure out where that's going to come from. And I think Microsoft got also seems like a great deal, you know, a, a deep relationship, the ability to integrate into their products, what they need and want to, and then also
36:09 Speaker 08 a share of the profits. So I think Microsoft fears not having a very strong consumer presence. That consumer beachhead does a lot for your brand over in the coming decades. And you've seen them kind of chipped away at that. They obviously have a gaming business, there's LinkedIn, but I, I unfortunately have the scars of covering their pursuit of Yahoo and also their pursuit of TikTok, many, many scars in this area. And so. I think you can see the opportunity that they saw in AI. I think, I think lastly too, and Satya was uncharacteristically public about saying this.
36:50 Speaker 08 If they can, by integrating AI into Bing, just make Google less profitable. That is also a win to challenge the business models of their rivals. This is also part of the agenda here. So it's very multifaceted and I think will evolve in a myriad of ways.
37:10 Speaker 04 One thing I haven't talked about on this series so far, you mentioned it, is the connection between cloud and servers and this tech. So how does Microsoft, Microsoft has this existing cloud business. How does that fit with their partnership slash investment with OpenAI? And how does that work for Google or Meta? Is there a underlying or secondary business that all these companies are in when it comes to sort of running these systems?
37:34 Speaker 08 You know, Microsoft has been locked in a war for companies to use its remote computing technology called Azure. It's a tight three-way battle between Amazon, Microsoft and Google. And Amazon's the clear winner. And so to land big companies and partnerships using your technology can not only be good for business, but also can help you literally refine and scale your systems to then open that to more people, to compete more for companies. So Microsoft can say, hey, work with us because you can have these cool AI features rolled in. So while, you know, OpenAI is not pay, the way their deal is structured is Microsoft's investment
38:17 Speaker 08 is essentially in the form of credits for Azure. So it's not like getting big checks cut from OpenAI directly.
38:24 Speaker 04 They're saying you can use our software for free slash not for free, but you earn $10 billion with the credits using to use our systems.
38:31 Speaker 08 Yeah, but what it means is they are then getting access to technology that they can resell and scale. So it's very much, again, in the war with Google and especially Amazon, a key strategy for them.
38:47 Speaker 04 So let's talk about business model. You mentioned this earlier. There's a couple of different theories. One is all the winning, all the winners are the companies we already know. It's the Googles and Microsofts of the world. There's a second, not necessarily competing theory. It says, and there'll be an OpenAI and Stability and a couple leaders that have sort of, you know, staked out a claim in AI already. And the third is, this is a platform. All these companies are gonna, there's gonna be a whole new slew of companies making stuff on top of this technology. They can be winners as well. And the example we always hear about is Uber developing on the iPhone. You hear different opinions. First of all, is there a conventional wisdom on this right now? Or is it up for debate from the smart folks you talk to?
39:31 Speaker 08 You know, increasingly, the smart folks I talk to seem bearish on startups' ability to make a lot of money here.
39:39 Speaker 04 Which is a bummer for some of the people who have already been on this podcast telling me about their new startup that's gonna bring a lawyer into the courtroom, or that's gonna help you coach you through loneliness. And you're saying a lot of ECs would be skeptical of this at this point?
39:51 Speaker 08 At this moment now, but I think the way this industry will evolve is like, we don't yet know what the, to use your analogy, Uber to the mobile phone, we don't necessarily know what the Ubers of AI are gonna be. And now, this is something crypto also said. So as I say it, I have to check myself. But I think that's kind of true. And so, the category of startup to be wary of as a business is, you know, there are so many open APIs and you can use, you know, OpenAI's technology and people are open sourcing all of this too. So, you know, very quickly now, two developers can come up with a new game, a new way of accomplishing some productivity task.
40:34 Speaker 04 Which doesn't, and that's a good thing, right? That's a good thing for them. It's the, the question is, is that something that an investor wants to put a ton of money into?
40:44 Speaker 08 Right. Well, A, it doesn't need a ton of money. And then B, you know, how do you ensure that isn't going to be eclipsed right away? Or how do you think about how that will compare to the open source version that someone else could release tomorrow? And so that's kind of what's being sorted through. I mean, one thing I hear a lot is the value will be in the data and the proprietary data sets. And so companies who will be able to make money will basically be running these algorithms over unique data sets. And that's where the uniqueness, I guess, will exist.
41:18 Speaker 04 Meaning I'm in healthcare. I have access to a lot of proprietary data that you literally can't get anywhere else. I have access to it and I'm going to integrate AI with that and create something that you can't duplicate.
41:29 Speaker 08 Right. And therefore my, you know, patient predicting what kind of treatment this patient needs, right? Because it's been trained on data no one else has, it's that much smarter. I mean, that intuitively makes sense to me. And I think like key data sets, AIs that are focused on very specific verticals, you know, sales prospecting tools that are using, you know, unique sales information. But you do start to see the moats around these business kind of crumble because what if you could just rely on the whole internet to come up with sales prospecting information, right? And then all of a sudden that sales lead list, right? That one seller has doesn't have much value.
42:10 Speaker 08 So, I mean, there used to be this joke that the business model for an AI company was an Aqua hire to Google, right? So every VC I spoke to who had invested in some AI company said, oh, my exit plan is Google's going to buy it because Google was hoarding researchers. And the people who said that now stop saying that because, but they don't know what else to say. And so I think we'll figure it out. I also think that to your point about whether this makes the incumbents stronger, I mean, absolutely, right? If you, sticking with sales, right? If Salesforce can very quickly deploy AI in a way that makes its products stickier, that's a huge win for Salesforce.
42:54 Speaker 08 The path to Salesforce being disrupted by a startup in this space is harder to see at this point, but it could just be that that's where we are in the cycle.
43:03 Speaker 04 When we first started talking, you said, you know, some of the optimism is turning a little bit. People are worried, have concern about the actual products, what they're going to do. I'm talking to you mid-April. The EU just basically said, hey, we're going to start trying to regulate this stuff. And most people I've talked to, especially at big companies like Microsoft and Google, say, regulation is very important, and we think this should be regulated, which is one wild, because big tech companies never used to say, we hope to be regulated. Now they say that out loud. Do you think they really expect to be regulated? Or is that a thing they say and assume that it will never show up, or by the time it does, they'll already have moved on? And secondarily, I'm very skeptical, particularly
43:47 Speaker 04 watching the US government try to grapple with social media for the last five years, that they're remotely going to be able or willing to grapple with technology that even the people who are using it don't understand right now. So I'm wondering what you think of sort of the new talking, what seems to be a new talking point from tech, saying, we welcome regulation, please come regulate AI. We think that's great. Do they mean it?
44:11 Speaker 08 Well, they mean, stop my competitor from doing something that's going to disadvantage me. Right? So I actually think that there is a genuine desire to see something, because it just feels like the Wild West. And I also think China and our kind of rivalry with China, particularly in the technology sector, is part of what's driving that. We had reported recently that venture funds backed by US investors were investing in Chinese AI startups. And, you know, it kind of prompted an outcry among a lot of investors in the US who were thinking, okay, you know, why are we emboldening rivals in AI, which many people classify
44:56 Speaker 08 as almost like, you know, weaponry, right, or secret technology in some way. So I think there are a lot of people in Silicon Valley who see a data point like that. They see, you know, how fast OpenAI is moving. And also, it's not so hard to imagine societal harm as it might be. I mean, in targeted ads, we could imagine societal harm of, you know, your discrimination and marketing and all these things. But AI is really a different beast. And so, yeah, it is.
45:29 Speaker 08 By no means clear how on earth a government will will dig its teeth into this, but I think you're you're starting to hear a lot more people bringing it up and traveling to Washington. And so I expect a lot of working groups soon.
45:43 Speaker 04 Yeah, it seems like something where maybe there'll be some legislation introduced in the next couple of years, but I can't imagine anything meaningful getting passed, at least in this country for a long time. But I'm always happy to be wrong. I want to just circle back and tie this up with a Gestalt question. And I asked you about the mood and people are excited. And we talked about how this comes on the heels of sort of the crypto flame out. There's an argument I've been hearing a little bit recently, kind of quietly, that it's not just crypto, that Silicon Valley really has not had a big hit in a long time. And you can play a little mind game where you try to go back five years and then you can keep moving the clock back and try to find a new company founded in
46:27 Speaker 04 X number amount of time. So five years for this argument that has been meaningful, that is really breaking through and you in the in the five year version, you can't find one. Is it possible that Silicon Valley has has gone sort of fallow for an extended period of time? And is that part of what we're seeing here, the the excitement around AI in particular? Or am I overstating it?
46:48 Speaker 08 I think it's hard to say. I mean, what I also think we've seen, of course, is that technology and innovation is really spreading throughout business. I mean, the number of new streaming subs has gone to incumbents and media, right? Like Disney versus on a relative basis Netflix is is sort of the the rest of the non tech world catching up to the tech world. And I think that has kind of defined definitely the last five years, you know, probably a bit more as well. But I don't get any sense. I mean, I still pay strong attention to like, where's the talent going? Where do the engineers right out of Stanford want to go work? And even with the disruption of the pandemic and so on, there's an incredible concentration of talent and capital coming out of tech. And
47:33 Speaker 08 I think not just in AI, but energy and batteries and climate. And there's just, there's a lot of activity. So I don't think the lesson is so much that the valley's lost its mojo. I think it's much more that. Technology and innovation are being impactful in in many new ways, too. But AI is a jolt of energy here. You know, there's some buzz forming. Our weekend section kind of helped coin the term cerebral valley, which is another crack at kind of giving.
48:04 Speaker 04 Tell us what cerebral valley is.
48:06 Speaker 08 Well, cerebral valley would kind of be like the AI, you know, especially in San Francisco, where a lot of AI companies are clustered, that area and that community, which as far as I can tell, has way too many happy hours in a given week. I just didn't don't know how it's possible to hit so many AI happy hours yet. Yet, you know, the youth out here are doing it. And so.
48:26 Speaker 04 Thought I thought I read in the information that no one drinks in Silicon Valley anymore. It's all it's all California sober.
48:31 Speaker 08 They're drinking their non alcoholic beer at their California sober AI happy hours. And, you know, cue the next season of HBO's Silicon Valley or Max's Silicon Valley, whatever it's called. But no, I think it's, I think we're going to see a lot that, you know, frankly, is going to bend our brains a little bit more than last generation of tech. So I think everyone should buckle up.
48:54 Speaker 04 I'm getting ready to have my brain bent. Thank you, Jessica Lessin.
48:57 Speaker 08 Thanks, Peter.
49:06 Speaker 04 Thanks to Renata Nyborg, Dora Berman and Jessica Lessin. Next week on the final episode of this AI mini series, we'll hear about the ethical pitfalls some people in AI are trying to navigate. And I also hear from Inside Google, which dominates tech today and is trying to figure out how to do that in an AI future. Thanks to Matt Fraschica for producing this episode. Meghav Kunnan is our editor. Shalani Carter is our engineer. Brandon McFarlane scored and mixed this episode. Talk to you next week.
49:48 Speaker 09 I work here. You know, that thing you have to wake up for every day? I know, I've made things a little difficult in the past. Okay, a lot difficult. But I got some help from UKG, and finally, I understand. Now I'm making work work better. I've even got AI agents helping me find new ways to help you, your people, and your business. Because when work works, everything works. UKG, HR, Pay, and Workforce Management.
50:19 Speaker 10 Everyone's got a hard question about AI. Your job, your kids, where it's going. Anthropic was built to surface those questions and share what it finds along the way. Because there's hope in hard questions. Ask yours at claud.ai/spotify and keep thinking. That's claud.ai/spotify.
Previous Exclusive: Google’s Sundar Pichai talks Search, AI, and dancing with Microsoft Next Microsoft CTO Kevin Scott on AI copilots, disagreeing with OpenAI, and Sydney making a comeback Page updated October 13, 2026 · transcript and summary are machine-generated