Is the AI Boom a Bubble?

0:00 This week we're joined by Mike Isaac. Mike is one of the best tech reporters to ever do it. He spent more than a decade at the New York Times and is the author of Superpumped, a book about Uber that was adapted into a TV show. Mike and us cover the same companies and technologies, but we come at them from pretty different angles. He has a very deep background in business reporting, and he's also very well sourced in the valley. We spent a lot of time in this episode talking about how the AI race looks right now from his perspective, the challenges in covering it, and also the reemergence of Uber's founder, Travis Kalanick. A note about this episode, we recorded this conversation a couple of weeks before Mike announced that he is leaving the New York Times and joining the Atlantic. So, if you're wondering why we didn't talk about that, he hasn't announced it yet, but congratulations to Mike. Mike, thank you so much for being on the podcast.

0:57 >> Thank you for having me. >> You have reported on Meta, previously Facebook, for I think like 20 years. I can't think of anyone better to get into about this recent settlement. Do you want to maybe just like talk about like what the rough outline of the settlement is? >> So, basically Meta, after years of court battles every level, the this one being with the attorneys general, many attorneys general across the United States and in DC, they decided to do this huge settlement. The upshot is they're paying 12 billion that could turn into 17 to 18 billion, and that's less important to me than basically making a series of product changes that would limit the amount of time that kids can spend on these apps, you know, whether it's a 2-hour time limit during the day or during school years and school hours they aren't able to receive push notifications, they have 15-minute check-ins, just like very drastic scaling back of how kids are allowed to use these apps. And I think, you know, the headlines usually the mainstream the mainstream media usually focuses on the money amount, but I think the product changes are where the rubber would hit the road here if enacted in a way that works and are not easily worked around.

2:19 Kids like if their ages are actually like locked in and they can't sort of like just skirt it and these things are enforced. That is the most consequential thing and Meta is of course reframing their capitulation as a as a we we love and care about children and want everyone to do this. >> Yeah, I I agree with the policy changes about how these apps and services actually work being the really big news. And obviously the first thing that came to mind as someone who has also covered social media and like the negative impact of social media for so long.

2:56 I feel like there was a cycle there for many years where every once in a while somebody would talk about how, well, you know, in China the rules around social media and gaming are very strict and the government forces these companies to limit the amount of time that kids can stay on social media and how much they can gain and on the weekend and if school time and all this stuff. And, you know, people have various reactions to that. Some people were like, well, that's authoritarian, that's against the people's like freedom of information and freedom of speech and so on. And then I remember it made like a really big news cycle obviously when like people told Joe Rogan about it on the podcast and he was like, well, you know, they're like the TikTok version in China is so much better for kids than the one we have in the US.

3:43 All of this in order to say, it's not the same, but it's like we're kind of ending up in like a more restricted social media for kids for the for the benefit of kids in a like in a rather roundabout way, but like not that different than China, I guess in in like in in in policy and in goals. It feels like a gigantic shift in how social media would operate both just like socially and then as as a business. Is it not apocalyptic for Facebook to enact these kind of limitations?

4:23 >> No, I I think you're exactly right. I do think this is a huge moment and it's funny we can or cannot get into this, but like there are a lot of groups that are actually protesting this for a number of different reasons. Making the China allusion that you would say, but for a different reason. But I do think they have been obsessed with and focused on getting young people into their apps forever because like people age out of social media and new companies and new ways of browsing the internet and browsing, you know, how to interact with other people on the internet do come around and I think TikTok is probably the best example for them. I remember covering the FTC trial in DC and they would have all this really great discovery around, you know, internal emails between Mark Zuckerberg and Sheryl Sandberg or whatever when TikTok was coming. And they were freaking out because TikTok's growth was huge and huge in particular with young people.

5:16 And so Meta is always in this position where they're like, "We are going to be unseated. We need to keep the hopper full of kids to like keep the next generations on here." And this can severely limit that sort of, you know, assembly line of keeping their users up, which are still growing by the way like across the world and you know in part because of their own maneuvering and ways to keep people coming back. But I do you it has the potential to be a {quote} "watershed moment" in how young people use social media. That said, like look, like there are a number of ways to get online now. YouTube, I think is actually probably under-covered by the mainstream media, but and when I say mainstream, I mean like you guys are very good at this, but I feel like big publications often overlook them and over-index, I would argue, on Meta sometimes because like so much stuff is on YouTube and so many young people are on YouTube all the time and TikTok, I think they've gotten. So, like the the question I have and the question that Meta is really asking right now is like what does that mean for the rest of the environment? If they aren't bound by, you know, settlements or agreements and certainly not regulation, are they going to make changes or not? I think Meta's making this very self-serving push to get every other company to agree to join them and it's so like, you know, they position it as we care about children, but I I mean it's very clearly like we don't want to be at a competitive disadvantage, but I think folks who hate Meta can still kind of be like okay, you have a point there, you know, if they can just go switch to another app very quickly. Does that make sense?

7:04 >> Totally, yeah. But there's like two tiers to the settlement. There's like, here's what we're willing to do alone, right? We'll pay whatever it is, 12 billion, and we'll limit the app by this much a child is allowed this many hours on the app and so on. But then, if like Snapchat and TikTok and everybody else signs on to the same terms, they'll pay more and they'll limit the app more. I'm not educated on how like the tobacco companies were regulated, but I've never heard of a a settlement being structured this way where it's like, we'll do more if you do more, which is is wild.

7:33 >> Totally. They they very very aggressively architected this with the AGs to make it like, if you're going to take our pound of flesh, we want it to be taken from everyone, and here are the incentives, and here are the incentives to get the AGs to put the put the screws to them, not just us, you know? Which I think will probably happen. Now, the states have direct financial incentive to push TikTok and YouTube very intensely, if that makes sense.

8:01 >> To this question of onboarding like the next generation of Meta users, I think that in popular culture, the image of Meta and and Facebook specifically is that it is a company and platform in decline. It's like people love to talk about how Facebook is no longer relevant and young people are no longer using it. Is that at all reflected in the company's financials and like what they're willing to say about users at this point?

8:35 >> I'm really torn on this. My theory, which I don't know if it's right or not, but I think, is that social networks as big and as entrenched as Meta is with its four major apps that serve basically half the world at this point, will take a long time to die, you know? Like, it's not the same way that MySpace sort of came and went. I think MySpace and some of its ilk from that period and then later like apps for smartphones that were social networks don't have the same sort of like tender hooks into all different parts of the internet that Meta does, you know? And I I do think it's accurate to say young people aren't using Facebook, the big blue app, Facebook proper, in the same way that previous generations did, at least traditional Facebook proper. By that, I mean like they're not scrolling the feed as much as like boomers are, basically, the folks that you guys write about that are consuming the slop that's on there. But, I think I should write about this at some point.

9:39 It's kind of been covered, but like young people love Facebook Marketplace, and they they are using that like a big time. Like, I've seen folks say like, "I wish Facebook Marketplace weren't attached to the rest of Facebook because it's genuinely useful. It's become a Craigslist competitor and replacement in a lot of markets." And young people are just on there sort of shopping the same way they might shop on like Depop or something. But, I think as a meme like Facebook is washed or Meta is washed and no one's using it. I think that's incorrect. Does that mean they're like on it for hours a day in the same way they might be on TikTok?

10:10 No, I don't think that's right. But, I do think Instagram Instagram is like the crown jewel of that company now in terms of social media traditional social media. WhatsApp and Messenger are very valuable in terms of like potential future monetization, and they still have enormous utility across the rest of the world. So, like yeah, I push back on the like Facebook is dead thing, but it is it is certainly bleeding in some ways and certainly has to worry about young people and this settlement doesn't help them in turning that around if that if that makes sense in a framing.

10:45 >> >> I mean, I am washed. I am chopped. >> Two washed chopped guys. >> Yeah, but it's like ironically Facebook or Meta owned platforms have never been more integral to my life because I'm a parent now and Facebook parenting groups like local Facebook parenting groups huge. And like so many critical things about my community happen on WhatsApp. I've never been a Facebook user. I've never been a big Instagram user. That that wasn't my thing, but it's like now more than ever it's it's really important.

11:19 So, I think you need to also look at like who is saying that it's over. It's probably people who were there and then left. They're like millennials. They're like 30 40, and they're not using it, but it's other people are using it in different ways. >> The groups thing, you're you're totally right. Like, this is it's become like a a utility in a lot of ways, you know? Like, it's just like, what do I need to go in and use it for? WhatsApp, I'm organizing stuff. Messenger, Facebook groups, I'm like, we have our PTA or whatever you know, we do. I think it's very much that for a lot of people.

11:49 And then, like, passive consumption and like entertainment is like an Instagram thing, basically, now. That's where a lot of people go for that. Those millennials, I would say, that have abandoned Facebook are watching recommendation videos on Instagram, if that makes sense. >> Sort of to transition us a little bit into AI, I think, again, as someone who is not I don't think I have the insight that you do into into Meta as a company, but I think you can tell a story about how, in their attempts to like make bets on the future, we've seen a lot of endeavors that flopped, right? It's like, there's the metaverse stuff, there was crypto, and you recently did some reporting about like betting apps, you know what I mean? And you can imagine you can imagine Zuckerberg being in a room being like, I want like this is huge, we want in on this. I think you can fairly say that they abandoned some of these projects. When you look at their very aggressive, expensive push into artificial intelligence, does that strike you as an endeavor of like a different scale or kind? Or like, is this more sincere or important?

12:57 Or is this just another like, well, this might be a thing that's big in the future, so we need to try it. >> I think there's two things actually going on here that maybe aren't mutually exclusive. On the one hand, Meta has actually been working on AI for a long time in different forms, right? They've had this research organization called FAIR internally, similar to Google buying DeepMind like forever ago, you know? Like, they all have used and developed AI. I do think it's almost trite now, but like the chat GPT moment in 2022 allowed the rest of the industry to take risks that they weren't willing to do before because of how big and closely scrutinized they are. So, like if you go back and look at when Microsoft released that what was it called? Tay or whatever chatbot that immediately started spouting slurs everywhere upon opening. They don't want that. They don't want to be like, "Here's, you know, Face- a racist bot or whatever." So, I think like they they as well Zuckerberg as well as Sundar over Google, they were like very skittish.

14:02 And then Sam Altman comes along, unleashes this bot that immediately gets a lot of traction, and every company's like, "Well, guess what? We have to like throw caution to the wind and like getting beat is way more existentially terrifying for us than like putting racist chatbots into the world, so we're going to change it." My criticism of him would be I think he's like I actually very much identify this. A millennial with ADD and like basically is like, "What is the latest thing?" and gets like easily distracted, but then obsessed. And I think also really closely tracks behavior online. I mean, it was what they they literally had a very clear window into it years ago when they bought the startup called Onavo, which was a way to track everyone's downloads and behaviors on their apps and what people were using.

14:51 And they had like a very clear window. Usually, it's pretty opaque and you have to do it out, you know, like sort of third-party firms, but this was like, "Here's what people are downloading and using." And so, what they would do is see that and clone it immediately. Now, they have like they still have ways of doing that, and I think but I think it's like to the to your point about the betting thing, like Mark sees stuff online bubbling up and is like "Make that ours now, you know." and I think I think his judgment is really in question on some of these just just in terms of the timing. Like betting apps are pretty much reviled by a lot of people now. I don't want to say universally reviled because I think a lot of people actually use them and there's a ton of people who are pushing for them, although I don't know how big that the number of people is. But like it's one thing if you're Calci and trying to position yourself as a responsible betting app, but Meta already has like this sort of noxious cloud over it and then saying we want to like do betting apps. Like it's just it's really it's hard for me to see. So I I would argue it's a bit of both. He kind of trend hops and I think he's been allowed to do that because their underlying apps still make a ton of money, but AI has been something they've been working on for more than a decade in some form at the company.

16:10 >> Also, to speak about the strategy, he's coming off like a decade where that that worked perfectly, right? They did buy Instagram, they did buy WhatsApp. And And those have paid off beautifully, like exactly as designed. It was a thing that was bubbling up and it fully exploded and they owned it, right? Like they got in at the right time. Actually, I mean, I don't know if you have anything interesting to say about this, but I do remember I was a very lowly blogger at the time, but I do remember that era.

16:36 No, you you were you were doing serious reporting at the time. Like you were I remember you were reporting like you were reporting on these acquisitions. Like you were doing like quarterly financial reporting, I think, at the time of the exact When you look back at that time when it's like they were on this I don't want to say monopolistic buying spree, but like this super aggressive, like anything that is coming up that could possibly challenge Facebook's popu- Like how did that seem at the time? You know what I mean? Like did that seem like, "Okay, it's normal. Like Face- Facebook is like this is what happens." Or were you like, "This is this is nuts." Like in retrospect, it seem it seems nuts.

17:10 >> This was like you said, like a totally different era. Google and Facebook were like the companies you sell to, basically, right? Every startup could be like, "All right, this is our eventual exit if we don't want to like keep going, just sell for parts or sell the whole thing. I don't think I would say I ever regret any of my coverage. I do think there are things I wish I was more critical on at the time, but I think the general eye towards covering Silicon Valley was certainly back east, but even out here was more like, who's up, who's down, who's going to win, and like, also not treating it with the like gravity of that we treated with now, right? And like, it's seen as more as like a story of look at these kids in their hoodies getting rich.

17:51 And like, I like I genuinely remember when when Instagram was bought, it was like, you bought an app for $1 billion? That's crazy, you know? Like, and now it's just like that's like a discount, basically. So, I I do think it was just a different way of looking at the world of technology, not just by the people covering it, but by a lot of the lot of the world, you know? And I think there's a moment where I certainly started getting smarter and more critical as my career progressed and as time sort of moved on. I wrote about this in my book, but like, I think the Trump getting elected was like this moment for a lot of folks as they ascribed tech and social media, which I think is a faulty assumption for a number of reasons. I think Trump people actually wanted to elect Trump, or a lot of the country, let's say, wanted to elect Trump. But like, it ascribed a level of power to tech that didn't exist in a lot of people's heads before that moment. And so, I'm grateful that now folks like you guys and and and, you know, some of the bigger orgs are like covering it more intensely, but it was just like a different moment, and acquisitions were it's under the Obama administration where they they were basically rubber-stamped and like rarely challenged, if ever.

19:11 >> Totally. Yeah, whereas today if like if you saw an headline suddenly that I don't know, meta acquired TikTok or something, people would be very upset. >> Totally. And also like Obama, you have to remember, Obama was pushing the internet and technology as this new way, this is the new economy, we have to get on board, we have to retrain people and get new skills because, you know, the old middle class and the old economy is not going to be the same and so information workers are really important. So, it like it did dovetail with his idea of what people should be thinking about for the future and I don't think that's a ridiculous notion, but it also had the sort of some would say unintended consequence of consolidating a lot of power in the hands of a few companies out here, which I think now they are and certainly Obama himself or like reckoning with, okay, what does that mean and how do we unwind some of this?

20:06 >> This podcast is brought to you by NordLayer browser. NordLayer browser is an enterprise browser that secures online work without getting in the way of productivity. It gives IT teams full visibility and control over every web-based AI tool, SaaS application and download across the organization while making sure employees can still enjoy a seamless browsing experience. Designed for today's distributed workforce, NordLayer browser helps companies protect access to their internal resources and prevent accidental data exposure. Go to nordlayer.com/browser/404podcast and use the coupon code 404podcast10 to get NordLayer browser at 10% off today.

20:45 I have a pretty simple test for whether a piece of clothing is worth buying. Does it quietly become one of the things that I wear every week? I feel like I could say that about so many Quince items from their 100% Pima cotton lux touch t-shirts, which I love to layer, to their stretch chore jackets and crew neck sweaters. Quince has become a real go-to for me and if you give them a chance, I think you'll have the same experience. Quince makes high-quality wardrobe staples with premium materials like 100% Mongolian cashmere, organic cotton, and merino wool. Their cashmere sweaters started at just $60, and everything is priced 50 to 80% less than similar brands because Quince works directly with ethical factories and cuts out the middleman.

21:24 Quince has slowly overtaken my wardrobe and my house. It's not just clothes. Their home goods are amazing, also. I have Quince towels, bed sheets, a Quince rug, and a Quince duvet cover. I haven't found any Quince products yet that I don't absolutely love. Find your next fall favorites at Quince. Download the Quince app for app-exclusive offers, or go to quince.com. Get free shipping on your order and 365-day returns. Now available in Canada and the UK, too. And when Quince asks where you heard about them, let them know it was the 404 Media podcast. It's the best way to support the show.

21:57 Holiday shopping is stressful enough if you're a buyer. But on the other side of every click is a business trying to get a mountain of orders packed and delivered on time during a time when carriers are often overwhelmed. During Black Friday and Cyber Monday, that pressure multiplies fast. ShipStation is an end-to-end order fulfillment platform that brings all your orders into one place, so your team isn't scrambling when volume spikes. It's automated rate shopping compares UPS, USPS, and FedEx rates in real time with savings of up to 90% off. We run the 404 Media merch store, and ShipStation makes it easy for us and our fulfillers to keep track of all of our orders, decide which carrier makes the most sense for our specific order, and get our products out the door without stress, because the holidays are stressful enough. Peak season doesn't wait, and neither should you. Get ahead of it today with ShipStation and get 60 days free at shipstation.com with code 404 Media. That's shipstation.com code 404 Media. Taxes and fees apply.

22:56 >> This episode is sponsored by BetterHelp. If you thought about trying therapy with our long-term partner, BetterHelp, insurance options may now be available nationwide with an average copay for eligible members of just $23 per session. October 10th, it's here. Maybe it's past, depending on when you're listening. It's World Mental Health Day. you know, stress does not politely wait for that venerated of holidays. Burnout does not RSVP. Your brain does not check the calendar before it starts running that circus and just you know, spiraling out as you're trying to sleep.

23:32 It's getting close to midnight. You've got to be up in a couple hours and then it's just going. It's just going. You know, you know, your your brain, it's not waiting around. Every day is Mental Health Day and it should be a part of the regular checkup and routine of your life. Therapy isn't just for these big moments and these big stressors. We're all going to have them, but you know, talking to a licensed therapist can help you work through the smaller things. Weird relationship dynamics, something that just kind of feels off in your life, it helps to have a professional to like sit down and talk to and BetterHelp can help you find that professional. So, if you've been waiting to start therapy, BetterHelp works with more than 125 health plan networks and eligible members have an average copay just $23 per session with insurance.

24:18 Don't have insurance or don't live in the United States, you can still get 10% off your first month. So, visit betterhelp.com/404media to explore your payment options. That's b e t t e r h e l p.com/404media. >> So, in the race between the biggest AI companies, there is this idea of a compute crunch where the AI companies, in order to train better models, they need a lot of compute. This is why we're seeing like this huge data center build-out. They also need the computer in order to like process all the requests that people have for these AI tools.

25:04 And if Meta is serious about AI, and if it wants to be one of these big companies that competing with OpenAI and Anthropic, how is it possible that, as you reported, they might lease a bunch of this compute to a competitor like Anthropic? Wouldn't that be against their interest to help a competitor develop better AI tools? >> This is all so weird and insular insular and circular. It goes to like xAI, for example, leasing to Anthropic and Google buying compute from xAI and like it's this I think the calculus is the hyperscalers that are building data centers are like overbuilding in some ways in a lot of ways, right? And so their demand for their products, like let's say Meta for instance, people aren't demanding the use of their AI products at least the direct consumer products to the degree that they need the amount of compute that they built.

26:13 They are using some of that a good amount of that for training their own models, but like it's sort of like there's a lot of the pie left over. So I do think there's some level of rationalization by these companies where they're like, "Okay, this stuff isn't sitting dormant or not moving or whatever or not computing. So we offload some of our cost by selling that capacity to competitors in the meantime, and then they write into these contracts we can terminate it with 60-day notice at any time, basically. So like it really is a weird the circularity the circularity, if that's a word, of all of this is just so strange and wrapped up in a level of frenemies and mutual aid, but also mutual destruction. I think we just have to like show how absurd and circular some of these deals are, and also really question the capabilities of these models, because as you were saying with like scaling laws, or I think that the the argument a lot of the time from Dario is like scaling laws are going to hold forever, and models are going to continue to improve, and it's all about compute constraints. And I don't know if that's true. I don't know if I believe that. Like, I think that is a self-serving argument in some ways, but I do think they very much believe that.

27:34 And so, it's sort of this thing where to keep the machine going, everyone has to like continue throwing in money and trading compute off, and then today you're a frenemy, tomorrow you're a an enemy, and we go cutthroat mode, basically. >> So, there's two things going on there, and one is, I mean, as you said, Face- Facebook basically overbuilds. They have this stuff lying around. I guess the math is like it's better to recoup some of the cost of this buildout, rather than, I don't know, denying Anthropic the compute to gain some sort of competitive edge. But then, I mean, there's the other thing that I obviously want to get into, which is I find very challenging as a as a reporter. I think all of us at 404 find very challenging, which is I am testing their products. I am looking at what people are doing with them, and I see the obvious gap between what is being promised and what's actually on offer. We've done this, you know, song and dance like a million times with tech companies before, and we're very focused on that, and I think that's fair. And then, there is a whole group of educated, smart people who will look at the market and tell you to your face that it's all a giant bubble. It's like the biggest bubbly bubble that has ever bubbled >> Mhm.

28:49 >> in in the world. And that doesn't that feels off to me by a lot also. Like that does not feel true at all. But I guess the naysayers version of the future is at the moment is like at the moment the computer is like heavily sub- or like the AI usage for like an average user or an enterprise client is really subsidized by the AI companies because they have all this money that they raised and eventually they'll run out of that money, investors will want their money back, and when people see like the true cost of using the AI tools, they'll stop using and the whole thing like collapses to to to zero.

29:29 That's the naysayers version of of of the future. What is OpenAI's version of the future? Like how does that not happen in their eyes? >> So it's funny. I've gone through like personal journeys of what I believe and don't believe over the past let's say three to four years on a lot of this stuff. I went from like all of this is fake to some of it's real to maybe it's all real to no, it's not. And like I think where I've netted out is I, like you, am very skeptical on some of the promises these companies are making. I cast some real aspersions on Anthropic. I'm less cynical about their motivations because I do think that they are true believers, like like cult-like believers in what they talk about.

30:14 >> Anthropic specifically or all of the >> Anthropic, not OpenAI. I would argue. I would say Anthropic specifically. Anthropic I just think that they they really believe in the machine god and that it's coming. I'm very skeptical on all these big promises, but I also don't think that AI has to be great in order to profoundly shape change the shape of how the world works. And by that I mean it can be really shitty and still have profound consequences on how we we interact with the world. So like you guys have done a lot of reporting on like just what flows through the feeds of people's Facebooks for instance, right?

30:46 Or public company CEOs who who answer to shareholders and really just have to look at how to cut costs all the time have no problem in some ways doing massive layoffs and and just like replacing those people with chatbots or or like replacing like freelance outside contractors that they have with like their own sort of in-house versions that are slightly less good or dramatically less good. I'm very skeptical of like the sky-high sort of promises that these folks deliver on, but whether they do or don't and I tend to think that they won't, it kind of doesn't matter because I think they've sold a lot of people on the promise or even the tools they're offering now are like good enough. And by good enough I can still mean like delivering a bunch of like really shitty like results or error-laden stuff. And like I think at the end of the day, what really bothers me and where I come back to is like there are folks out there, you know, very prominent people who cast all of this as a sense of inevitability that guess what, this is coming whether you like it or not. And I don't agree with that. I think human beings have agency and the ability to choose whether or not we want this to come and I think there are a lot of people who don't want it to come and are pushing back on that.

32:03 My point of view is just to like make criticisms that are well-founded and and try not to be an equal sort of prognosticator in knowing what the future's going to hold. I just I don't trust anyone who believes they know the future if that makes sense. >> Yeah, yeah, that's fair. I guess very quickly two tangents. One is the the doom and gloom version of it or like the bubble version of it is not that similar dissimilar than the story of Uber I think in the sense that it's like the story that people would tell is that the cost of Uber rides is heavily subsidized by investors and that is a kind of a land grab, but then when they can no longer subsidize it, it's like the whole thing will collapse and like self-driving will never come and and and it's like the the whole idea doesn't work and it's like did it work? Did it work or not work? Like you Uber's obviously still around. So to me it seems like that did that did work.

33:02 >> Actually, I think self-driving is like a really great example to use because you know, we were covering it back then and it was always 5-year timelines. Everything is going to be autonomous in 5 years and this is why we can raise this crazy amount of money. The reason Uber worked is because they basically sold off all of their very expensive development including self-driving and Dara came in and became a cost cutter basically. I think what I would say is like the the thing that bothers me about Silicon Valley is that being wrong has no real cost to it because if you're wrong, you can just keep kicking the can down the road to the next thing. Let's look at crypto.

33:38 Let's look The metaverse is like my favorite example because it was so obviously in a lot of ways around the thing that I was just talking about of like get on board cuz it's inevitable. Like it's not inevitable. People like decided they don't like it and so it didn't become a thing, you know? but that's why I just like I think they're very helpful use cases in in in in the lens by using a lens to view like these statements from the people at the top of these companies.

34:07 >> The other tangent I wanted to to to go on is you guys reported on the OpenAI agentic hacking of Hugging Face. And in terms of like trying to stay measured in your reporting, we had a conversation about this yesterday. Like we had a meeting. We didn't cover it. There were many reasons. The primary one is that like there's only so many things we can cover, but it's like one reason that made me like turn away. Like I don't even want to like look at this is like if I started to write about it, you sound crazy. Like when you when you start putting down sentences about like just describing what happened, you start to sound like a crazy AI person. Like you have to say like over 100 AI agents escaped containment and coordinated attack on a company on the internet. It just sounds like you're like hysterical or something or like you're people get very mad at us when they think we're endowing AI with like intent and personality and stuff like that. But it's hard it's hard to talk about this stuff without saying sentences like that. And I was wondering like have you guys had conversations about that? Like have you had difficulty in like well, how do we describe this crazy thing in a way that is level-headed and doesn't make people think like we're talking about science fiction or something like that.

35:23 >> I would agree with your audience in that like describing personhood to these things is dangerous. >> Agreed. To be clear. >> Yeah, yeah. The way I try to reframe it a lot is that these are programs executing commands, you know? And like they they were told, they meaning these programs, were told to perform a task and those programs successfully performed a task within within the parameters they were set but also by not listening to like they had internet access. In this particular case, they were sandboxed to a degree but still had internet access and they were said, "Hey, go like break into whatever some company." And they did and like they had the option of continuing to use the internet and did. I think it is seductive to ascribe that level of like personhood and like, "Okay, the bots like convened and they started talking. How are we going to break out of the box?"

36:11 So like in 2019 actually, this is not to pat myself on the back at all, but in 2019 I went to a colleague Cade Metz. We were talking about AI in the context of was then Facebook and I was like, "You know, I've been thinking about like language and AI and like one thing that makes sense to me with AI, it's probably easier to catch things that are very concrete like numbers or particular photos or like there's there's certain things a computer has very good ability to recognize and prohibit, right? But I think language to me in my mind, to my non-AI researcher mind, seems much more fluid and ever-changing and difficult to pin down and seems just like a harder AI problem and Cade was like, "Hey, guess what? You're right. Let's go do this thing." And we wrote a story and like it's very outdated now, but like it it is a helpful use case for me to understand why robots are so good at coding now and are so flawed at I would argue good at coding because it's very specific parameters around like how computer languages work and you can build a type of program based within those parameters. Now, there's like, you know, coders would say, you know, I don't know, it made these certain choices and there's some level of like artfulness that coders can make, but like it's helpful for me in understanding hacking scenarios or if you give it a task, it goes and completes the task. That is historically what computers do. And so, yeah, but I agree with you like I think it's I think it's smart to be wary of how you write about it if that makes sense and it can be difficult. That That's probably why I also have not written about it.

37:52 >> I'm genuinely asking for your help. This is something I always struggled with. I think the next big step in like the financial journey for these AI companies is initial public offering, right? Like that is like the next huge thing that will happen and it seems I don't know, maybe soon. Like Anthropic, I believe, is in the process. He reported that OpenAI considered doing it this year, but then maybe pushed to 20 to to next year.

38:21 I don't understand like what are they timing for? Like why would Open AI want to do it next year? What happened that it that they decided to push it back? Like what what is the ideal time for these companies to IPO? Cuz in theory you IPO and then you you get the surge of investment and you have more money that they apparently really want to pay for all this compute and data center build out and so on.

38:46 >> Earlier this year, there were a few things going on. I have maintained and I think continued to be right that Open AI was in no shape to go public this year financially. Their books are in shape. They couldn't make it. But, a few things happened earlier this year and late last year. Claude Code changed its model last year and it becomes the coding god and their revenue goes through the roof because every coder is like, "Oh my god, I'm out of a job now." That drastically accelerates Anthropic's timeline, fundraising abilities, valuation, timeline for IPO. So, they become a real player. Open AI starts freaking out. SpaceX also decides to go public mid-year. Companies go public for two reasons. One, to raise an enormous amount of money, which will help them in their computing costs, which they need, and two, to cash out shareholders. And they have have they have a lot of shareholders that want to cash out at this point. But, I think primarily being a public company is a pain in the ass because you have to answer to the market every quarter. So, they don't want to, but they sort of have to. Open AI sees these things happen and they leak or someone there leaks or someone in their world, let's say, leaks to the journal earlier this year that they have plans to go public this year.

40:01 I don't think that was ever really going to happen, in my opinion. I think it was very aggressive and meant to like freak people out. And also, they didn't want all the money to go to SpaceX in their public offering because being first to an IPO can have its advantages if you want the market to allocate their dollars to you versus Anthropic or OpenAI. Now we're in this position where Anthropic is speeding towards an IPO cuz their books are good. They are still increasing in revenue run rate. That doesn't mean that they're profitable, but it does mean that their revenue is going up. And SpaceX was a flop, but OpenAI now is very I'm fairly confident it's going to be next year. They literally replaced Greg Brockman, the president, is now back in and like tearing up. He fired or he gently ushered the woman who was leading sales, their CRO, chief revenue officer, to the exit. Her entire team is basically following her out. He's rebuilding and like tearing things up and trying to like install this thing.

41:03 And their books are just not in shape to do this IPO. They all sort of need to go public to cash out shareholders, but really to raise money for compute. That's all they really gain from an IPO. >> And you're trying to balance getting there early, so you're you're getting the money, and that but like that's versus when you go public, you have to open up your books. And you when when you open up the books, you want it to look as best as possible. That Is that basically the consideration?

41:31 >> Yeah, one thing about IPOs, first of all, I should back up and say the other thing is timing the market, which is sort of a fool's errand. Markets change all the time. It's all It's an impossible task, but you try as best as you can. They, meaning the bankers for these companies, believe that the markets are in a better position. At the same time, like a war could happen. A trade embargo could happen. Like it's all very difficult. And then, exactly what you said, your books need to look good. And what they need to show bankers, and they need to show retail investors, is a growth story because it's a moment to cash out your existing shareholders, but what you need to show is that your company's best financial days are ahead of it, not behind it. And so the numbers need to be going in a positive trending direction. But, doing that means a lot of fiscal cleaning up internally, which takes a lot of work, especially when you've had years to spend prolifically with no consequence and not have to to answer to anyone, basically.

42:30 >> When companies go public, they have to file this form called S-1, which is like a disclosure form, which is when you're a huge company like that is really huge. And I don't do financial reporting really anymore, but I if it's a big company that goes public, the first thing I'll do is I'll pull up the S-1 and go to risk factors. >> Yes, that's a that's that's the key move. That's right. Yeah. >> And I can't wait to see the risk factors for the cuz it's got to be nuts. It's got to be like, >> >> you know what I mean? Like, oh, you know, and one risk factor is that it's like it will end humanity or something like that, right?

43:04 >> >> Yeah, yeah, exactly. One risk factor is the robot nukes us. >> Right, cuz it's like they are talking about that like seriously when you talk to them. Like, maybe it's a cynical, self-serving thing, but it's like, I don't know. Wouldn't you have to put it in the risk factors as well as if you mean it? >> Honestly, that's a great I hope you pre-write that story and are ready to push the button, cuz that is a brilliant angle, honestly. But, yeah, you're exactly right. The S-1 is sort of like a Christmas for for reporters, because you get to see for the first time all of the inside stuff.

43:34 >> Last thing about AI I've been wondering, I know you're a long-time Bay Area resident. >> Yes, sir. >> Bay Area lover, fair to say? >> I I I've I'm I was raised in Texas, but I am spiritually Californian and spiritually San Francisco Bay Area in particular at this point. >> Okay. Same. Same, also. And I'm an exile right now, but I do I do really love San Francisco and lived there for 8 years. >> Right on.

43:59 >> And I moved there in '07 and left in 2015. So, I really got to see the city change because of like that boom of like mobile and social. I know it's really crazy. Like you definitely saw it and felt it. And I've been wondering, do you feel and see the AI money in the city or is it or is it just like in the same continuum of like what was already happening? >> That period that you were talking about exactly. I got here in 2004 and so that was like post dot com crash and rebuilding and that was like pretty normalish, you know, or like normal on the continuum of San Francisco, but like things absolutely accelerated post iPhone and web 2.0, Facebook, Twitter, continues of Google and it jumped again post Uber, Airbnb, that wave of companies. The other thing is we moved out of the city from in 2021 to come to Oakland and I feel like more of a The East Bay is way more chill than the city in terms of that change you're talking about and the change in the city is like on steroids.

45:06 There's still like pockets of quote unquote normal see in like the old San Francisco, but the but like landlords are absolutely finding ways to like get people out so they can jack up the rent a zillionfold for new tenants who have like unlimited money and can come in like double asking price or whatever and it it it does bum me out because it it look like I'm not going to tell anyone not to live somewhere, but the reason this city was so vibrant and lovely is because it was not it was not as much of a company town as it feels now. It's always been a company townish, but you could still get flavors of different folks. There were artists, there were musicians, like they would work like the the bar and restaurant and coffee shop jobs and then go like have a cool part of San Francisco. You go to Slim's or the Parkside or the Bottom of the Hill and like two of those are gone and one's going to be gone by the end of the year, you know, like that is what I really worry about about around here, but I my my tip is Oakland is where it's at and I'm hoping we can hang on to that at least over here.

46:18 >> Has the AI money like impacted the city and and the area? Do you see that? >> No, for sure. The amount of money that people are about to make from these IPOs is is exponentially more than the last. >> I mean it already is. >> Yeah, they're already cashing out in what are called secondaries and like tender offers and stuff. So like there is already and that's before and that's after the salaries that they make which are also insane, you know.

46:42 >> Yeah. Yeah. okay, last thing in case people don't know, Mike wrote a book about Uber called Superpumped which became a TV show as well. I would say, I think it's fair to say that the CEO of Uber is is the main character. >> Yeah. Yeah, totally. >> And a lot of controversy around like discrimination and and at the company Travis was pushed out which by the way, I have to admit I didn't read your book and the reason is I remember when we got it at the office, we were still at Vice and I opened it and I read a few pages and it was like it reads like you're in the boardroom and and like I was like literally upset.

47:23 I was like how did he do this? And I was like >> Oh, that's that is the hugest compliment I've ever gotten for someone not reading my book. I actually appreciate that. >> It's like a minute by minute of what is happening. It's really really crazy reporting. Anyway, Travis gets pushed out, very controversial. Like when you think about like aggressive overstepping you know, tech company CEO, he's like the perfect example of that guy. Goes away. How long is it now? I don't know.

47:51 Like many years. >> Yeah, 2017 so almost a decade later basically. >> It's like people knew that he was working on this thing called Cloud Kitchens, which is like a ghost kitchen company, I I assume, but maybe other things as well. But recently he's popped up and specifically I saw him on a podcast called TBPN. And they were asking him like, "Hey, how come how come you're in the media again?" And he said, "Well, the media has changed. Like I couldn't talk to the media because it was so negative. And now there are all these other places where I can go and talk to people, notably this podcast that's now owned by Open AI.

48:26 >> Yeah, I should I should mention. Yeah. >> Yeah. and I'm I guess what A, what are you wondering about like the reappearance of of Travis? And then also his perspective on the media being, I guess, more favorable to him, less hostile to him. >> It's very funny. I got like literally a dozen people texted me that clip of or a clip was circulating of him and he said the words like The New York Times like a dozen times when >> Oh, he's talking about >> I will take it as synonymous for my name and I appreciate he didn't use my name, but also like a lot of people at the paper wrote about it, so it wasn't just me. But I had like a genuinely that was one of the hardest periods of my life just in terms of reporting and like stress and like my home life and it was very intense. It was like a year of intense stuff and that's not to like get garner sympathy cuz like it's that was a crazy year, but it was more like whenever I go back to that, like I have like a trauma response or I'M LIKE, "UGH."

49:27 SO LIKE people were sending me it and I was like, "I can't watch this. I can't think about it. I can't. I'm just It's like I remember how insane that was." That book and that reporting also came it was right before the Weinstein story and the Me Too movement and it was like really in the continuum of what I think was like this social moment of upheaval around bad behavior and misogyny or racism or company cultures or executives abusing power.

49:55 And so I think Travis like Travis did not do what Harvey Weinstein did, but I but like I think that was the sort of environment in which all of this sort of appeared. And now it it it feels like a very specific moment of time that is almost quaint in how things are are the environment we're in now. >> I just want to remind people like Uber was a huge part of like the Me Too thing, especially as it like that's how it was like a main link to like the tech industry. There was this whole story about the jackets. They got a bunch of leather jackets. Only the men got the jackets. The women didn't get the jackets. Like this is this is kind of what is happening at the time, yeah.

50:37 >> Totally. And now the in tech parlance, the Overton window for what people find acceptable or not acceptable has shifted so dramatically, a lot of this stuff doesn't get flinched at. And you know, like the Trump administration and like the sort of norms that they have established for how to act in the world or how to act online or how to act in companies have really dramatically shifted things. And I think And that dovetails with a lot of executives in Silicon Valley being like what quote unquote woke is dead and you know, get that woke BS out of here and we're going to like be ourselves along with the rise of these independent podcasters who have a more tech positive perspective on the world.

51:24 And so like it's Travis, but it's also Zuckerberg or the CEO Coinbase who's out there calling like it's actually very funny how obsessed with mainstream media they are even though they say we're all dead all the time, which is like a very obvious tell to me that they really crave the approval of this coverage and can't have it. So we're trying to create their own sort of thing. I think to they would talk to me tomorrow if I was ready to write a glowing story about them, but I'm not.

51:53 This is the other thing. Their audiences are welcome to go find that level of coverage and and perspective in all of these podcasts. I think that's great. And if that's where the CEOs want to go talk, like that's awesome. But like I I feel that my function and the function of some of the folks like yourself or like other organizations is a different thing in some ways or at least the type of stories that I do, that doesn't preclude other types of stories at our paper, which is huge. But like what I do is going to be a different thing. And I think that's fine, too, you know? And like I also, going back to what we talked about earlier in the episode, we're in a different way of viewing what the power of technology can be and I would argue it's worthy of scrutiny on par with how we scrutinize regulation in Washington and the political arena or Wall Street, right? Wall Street has been heavily scrutinized for a long time in areas of reporting. So like I'm used to it at this point and it's not going away, but near my, at least not not yet.

52:56 >> Will you watch a Mark Zuckerberg do 3 hours on like Lex Friedman or whatever? >> I can't I can't do these like 3-hour pod it's too much, man. Like that is like I don't use AI tools in my daily life, but if there's a way that AI can condense that for me, like maybe. But I don't know. It's just I'm not listening. I'll let I'll let my colleagues do that. >> >> I think also the media training caught up. Like Mark Zuckerberg can sit for 3 hours and like not slip, which is impressive.

53:23 >> He's way better at it than he was 10-15 years ago. I agree. >> Mike, thank you so much for being on the pod. Really appreciate you taking the time and your reporting as as always. Should people follow you anywhere other than nytimes.com? >> Your audience probably thinks I'm evil for still being on X, but I am on Twitter is what I call it and it's unfortunately it is where all of tech people still are so there's a level of having to be attuned to it but also like it actually is like a very useful reporting tool because literally all they do is talk to each other on there all day and certainly AI people. So if you hold your nose you can find me there. If you want to read the book it's still out there and thank you. It's an honor to be on a huge 404 fan and we got to get back and start gaming again soon.

54:11 Let's play Call of Duty with me. >> >> Okay, Mike thank you. As a reminder, 404 Media is journalist founded and supported by subscribers. If you wish to subscribe to 404 Media and directly support our work, please go to 404media.co. You'll get unlimited access to our articles and an ad-free version of this podcast. You'll also get to listen to subscribers only sections where we talk about a bonus story each week. This podcast is produced by Alyssa Medcalf. Another way to support us is by leaving a five-star rating and review for the podcast.

54:47 That stuff really helps us out. Please do it if you have the time. This has been 404 Media. We'll see you again next time.