Pod Insights Decoder Inside the players and politics of the AI industry About this episode We’ve got a special episode of the show today – I was traveling last week, so Verge deputy editor Alex Heath and our new senior AI reporter Kylie Robison are filling in for me, with a very different kind of episode about AI. We talk a lot about AI in a broad sense on Decoder — it comes up in basically every single interview I do these days. But we don’t spend a ton of time on the day-to-day happenings of the AI industry itself.
So we thought it would be a good idea to take a beat and have Alex and Kylie actually break down the modern AI boom as it exists today: The companies you need to know, the most important news of the last few months, and what it’s actually like to be fully immersed in this industry every single day.
Links:
Google defends AI search results after they told us to put glue on pizza | The Verge
Apple is putting ChatGPT in Siri for free later this year | The Verge
AI will make money sooner than you’d think, says Cohere CEO Aidan Gomez | Decoder
Humane is looking for a buyer after the AI Pin’s underwhelming debut | The Verge
2024 is a year of reckoning for AI | The Verge
OpenAI researcher who resigned over safety concerns joins Anthropic | The Verge
Hugging Face is sharing $10M worth of compute to beat big AI companies | The Verge
The AI drama is heating up | Command Line
Google and OpenAI are racing to rewire the internet | Command Line
Elon Musk’s xAI raises $6 billion to fund its race against ChatGPT | The Verge
Credits:
Decoder is a production of The Verge, and part of the Vox Media Podcast Network.
Today’s episode was produced by Kate Cox and Nick Statt and was edited by Callie Wright.
The Decoder music is by Breakmaster Cylinder.
Learn more about your ad choices. Visit podcastchoices.com/adchoices
Transcript 8,634 words · 10 speakers
0:00 Speaker 00 Company spend can be complicated. Managing it shouldn't be. AMEX Corporate connects cards, expenses, and payments all in one place. That's less time managing work and more time doing work that matters. Built to simplify, designed to scale. This is AMEX Corporate. Get started at go.amex slash AMEX Corporate. Terms apply.
0:28 Speaker 01 This episode is brought to you by Subaru. You know that place your mind always wanders to? The lake, the mountains, the homes of the people you love most? The completely redesigned 2026 Subaru Outback is built to help you get there. With legendary capabilities, standard symmetrical all wheel drive, and the confidence to go further. The 2026 Subaru Outback. Love never looked so good. Visit Subaru.com slash Outback to learn more.
0:55 Speaker 02 Hello and welcome to Decoder. I'm Neil Appert, editor in chief of The Verge, and Decoder is my show about big ideas and other problems. We've got a special episode of the show today. I was traveling last week, so Verge deputy editor Alex Heath and our new senior AI reporter Kylie Robinson filled in for me with a very different kind of episode about AI. We talk a lot about AI in a broad sense on Decoder. It comes up in basically every single interview I do these days. And then we also spend time diving deep on very specific problems that AI is causing with things like copyright law, misinformation, and all the other various policy challenges. But we don't spend a lot of time on the day-to-day happenings of the AI industry itself. And that's for a good reason. It's all moving so rapidly and changing so often that it's nearly impossible for the average person to keep up. Making sense of all the different players, who's building what, and what people in the
1:40 Speaker 02 industry actually think about where it's all going, yeah, it's a lot. So we thought it'd be a good idea to take a beat and have Alex and Kylie actually break down the modern AI boom as it exists today. The companies you need to know, the most important news the last few months, and what it's actually like to be fully immersed in this industry every single day.
1:59 Speaker 03 You'll hear Alex and Kylie talk about OpenAI and the gravitational pull it's had on the entire tech sector and how Google is trying to balance that out to mixed results. But they also got into the relationship between the overzealous hype driving AI funding and product development and the very real paranoia in the Bay Area around what's called P doom, which is shorthand for the probability that AI might actually kill us all. They also break down all the most important AI companies and how likely it is that some small AI startup unseats ChatGPT, as well as the open versus closed source debate. And as Alex puts it, the Wall Street-ification of AI. Oh, and look out for a mention of an effective accelerationist rave where Grimes DJed for the CTO of OpenAI. Those are all
2:41 Speaker 03 real words in the same sentence altogether. That's The Verge. That's Decoder. Okay, inside the actual business of the AI boom. Here we go.
3:00 Speaker 03 All right, Kylie, welcome to Decoder.
3:03 Speaker 04 Thank you for having me.
3:05 Speaker 03 Oh, I'm happy to have you on Neelai's behalf and excited to chat. You have been with us at The Verge for, gosh, just over a month now, is that right?
3:15 Speaker 04 Yes.
3:15 Speaker 03 Which is an eternity in the AI world. There's been a lot going on. We kind of threw you right into the fire and you've been killing it. You know, we have a pretty tech savvy audience on this show, but there's probably a lot of people that can sympathize with your experience, actually, of just jumping into this beat like this, especially right now. It's so frenetic. We're gonna go through and talk about some of the big themes and the big players and try to make sense of all of this as best we can in under an hour. But I'd love to actually start with you approaching this beat and what you're excited about in AI as you're reporting at The Verge.
3:52 Speaker 04 Yeah, that's a really good question because the excitement is difficult to talk about. It feels like we get a lot of doom and gloom. I was just blogging about this. Whenever I go on live news, they're like, is AI gonna destroy us and take all our jobs? Is usually what I'm asked. What a ride and so much drama. I think putting away all the OpenAI drama and looking at how the technology advances, I think is very exciting. I do like to imagine this idealistic future where artificial intelligence can detect diseases quicker, organize my emails and do all sorts of things to help us as humans and not really impede humanity. So
4:33 Speaker 04 that's what I'm excited about. I like this newsletter called Deep View and recently they had talked about how these researchers in Australia were using artificial intelligence to detect wildfires easier, which obviously is a huge problem over there. And I thought that's really neat. But yeah, it's been a lot to consume all at once in the first month.
4:55 Speaker 03 Yeah, it's like drinking from the fire hose every day.
4:58 Speaker 04 Right.
4:58 Speaker 03 It feels like. And you're based in San Francisco and you've lived in the Bay Area your whole life practically, right?
5:04 Speaker 04 Yes.
5:04 Speaker 03 I'd love to know what it's like being there and covering this stuff because, you know, having covered tech for a while, it really felt like the center of gravity was moving away from SF, especially going into the pandemic. And there were all the articles about SF is doomed, it's over. And it really feels like everything has moved back there. And really, thanks to AI, I mean, there's even like Cerebral Valley, right? As I don't know if anyone who lives there actually calls it that, but at least externally, there's this neighborhood where all the startups are and OpenAI and everyone's there, of course. But I don't know, just having gone up there and hanging out with you there a few times for trips in the last month or so, it really feels like the energy is high and it's all kind of buzzing with AI stuff. Is that accurate?
5:46 Speaker 04 Yes, I can't imagine doing this job anywhere else. I mean, of course you can, and there's plenty of talented reporters who do do that. But for me, it's very helpful. I live next to the quote unquote Cerebral Valley. I live very close to there.
6:01 Speaker 03 Oh, you do?
6:01 Speaker 04 They just got a Trader Joe's, which was huge news for all of us. So I take a lot of meetings over there and it's always AI meetings and they always make the Cerebral Valley joke. Only jokes. I don't see anyone saying that in actual conversation. It's Hays Valley for the uninitiated. But yeah, it's It's really easy to bop around to meetings and meet new startup founders who have slapped AI on their startup name. It's very helpful for going to parties. I went to OpenAI Dev Day, which we did not know was like gonna become a red wedding situation after. So I went there and afterwards I went to this rave. I was invited by another journalist. I went to this rave that
6:44 Speaker 04 was DJed by Grimes and I ran into the CTO of OpenAI, Mira Maroti, there. That is just like the kind of story I love telling. So I'm like, where else on the planet is that reality? That was crazy. So yeah, it's very, it's very helpful.
7:00 Speaker 03 AI raves. Okay, we need 10,000 words on AI raves.
7:04 Speaker 04 It was an EAC rave, actually. I didn't know before going, but it was an EAC rave.
7:08 Speaker 03 Okay, actually, I was gonna ask the big picture question, but maybe let's start there with these terms. So EAC, explain EAC to people.
7:16 Speaker 04 Yeah, effective accelerationism. I'm glad I got that right on the first try. Those are people, like at this rave specifically, they have like, don't tread on me flags, but they're basically saying, don't impede on our AI progress. We want to go fast. We want to go hard. We want to break things. This is imperative for the future of humanity is if we build this as quickly and aggressively as possible. So that is kind of a faction, right? I ask this question a lot to people who are, you know, actually working on AI. I think really all of that is just noise. I don't think we have to put like a whole huge emphasis on it. It's easier to see it in black and white. There's people who want to slow down and there's people
7:58 Speaker 04 who want to speed up. But I think, you know, that's mostly just people partying and tweeting. So in that order, I'll explain D cells, which is the opposite. They want to deaccelerate. They want to pause. They think that we're moving too fast and we should move forward more thoughtfully. And these EAC people can also call them doomers, which is people who think that something called P doom, their P doom is really high. They think humanity is doomed. So we should...
8:26 Speaker 03 P doom.
8:26 Speaker 04 P doom. You haven't heard P doom?
8:28 Speaker 03 I don't think I've heard P doom. So explain, what is the P in P doom?
8:32 Speaker 04 It's just people who think that we're gonna get killed by AI. And how high, how likely is that?
8:37 Speaker 03 What is your P doom?
8:39 Speaker 04 It literally changes so often. When I was going into this beat, this is actually something I told you. I said, you know, I'm really excited about the AI beat because as nerve wracking as it is to completely change beats, I was covering Twitter before. I really enjoy technology doing things for humanity that are good, obviously. Like technological advancement is a good thing. So I really like exploring how it impacts humanity and how humans are building cool shit with it. And then I started covering it for the past month and it has made me a bit of a doomer.
9:12 Speaker 03 Really?
9:13 Speaker 04 Mostly because I feel like so much of it is, and I'm blogging this currently, so much of this is kind of a cash grab, it feels like, and to impress investors. And it's not so much for the consumer as it is for money sometimes, it feels like. And I'm wondering, like, are the consumers being left behind? What about data privacy practices? And like, all of these ambitious goals to protect us, what happens to us? Like, for AI overviews, it's a very silly one. But them telling us to put glue on our pizza, I mean, it's just, what are we doing? It's just AI for AI at this point. So sometimes it makes me a little jaded.
9:54 Speaker 03 And for those who don't know, AI overviews are the new Google AI search results, which you wrote about, that told people to confidently put glue on pizza. We will get to that for sure. I want to folks stay kind of on the big picture, though, before we get into some of the players. So you perfectly teed this up. You have become a little cynical, which is, I think, the natural arc of any journalist. But you're realizing that as you are more getting into this beat, that there's just a lot of financial engineering going on, a lot of money behind all this naturally. And that influences what we see and what consumers see in the products. A question I've been wrestling with, and I think everyone has been wrestling with, especially in the last year or so, is it's pretty clear that generative
10:35 Speaker 03 AI is being stuffed into every product imaginable. It's kind of silly at this point where it's showing up. But the question I have is, why? And is it purely money?
10:49 Speaker 03 Is it because companies are seeing new unlocks that they wouldn't have seen before? Is it something else? I think it may be a mix of all that. But as you talk to folks and meet people at startups and the big companies and all of that, are you landing on or circling around the why of what is driving just the mania around all this right now?
11:12 Speaker 04 Overzealous tech executives and product managers is my guess. As you were talking about this, I was remembering Salesforce having like NFT projects. And I'm like, it's just, you know, it's just another acronym to tack on to things now and to drive sales and look cool. But if I'll be positive, I mean, some of the features Apple launched or is launching with the iPhone 15 and above, the AI features, I think are really useful and really cool. I want my iPhone to be more useful. As much as I'm worried about sort of the data privacy aspects, just practically, I would like to just tell my iPhone to do something and it does it.
11:51 Speaker 04 So yeah, I think there are good use cases and there are like good intentions, but sometimes it's just slapping it on to something to slap it on to something.
11:58 Speaker 03 Yeah, and you brought up crypto. I see a lot of parallels in the coverage and the hype to the last crypto boom in terms of a lot of companies tacking on AI, just to tack on AI to get higher evaluations and more press. That's happening, absolutely. It does feel different, though, in the sense that when I go into a coffee shop, either in LA where I live, or SF, or New York, or wherever, I'm not surprised to see ChatGPT open on a laptop. And before the, with the crypto boom, no one was actually using this stuff. And that's what feels different about generative AI is, you know, ChatGPT was, I think, the fastest software product ever to 100 million users.
12:40 Speaker 03 The revenue growth that Nvidia has seen is literally unprecedented for a company that size. So there is something real happening. But I think where I'm starting to have concern is that the hype train is maybe getting ahead of the actual value that's happening. Do you agree with that?
12:57 Speaker 04 Yes, so this is actually a question I asked everyone because I had to stake my whole career on it. I have to move beats and cover AI and will it just be like crypto? And will it fizzle out? And I think I even asked Neelai in the interview process because there's a decoder, I believe with Sarah, where he says, you know, like, the AI bubble's gonna burst, it's gonna be over soon. And I was like, okay, why should I cover this? If you think that, and we talked about it. But I do think the difference between crypto and AI is we see like actual proper use cases out of AI. The hype that is similar is that, yes, it's like, oh, you know, crypto is gonna change the world, it's gonna help these, you know, struggling economies, it's gonna help people be independent. And we didn't really
13:42 Speaker 04 see that at scale. With AI, people are using it for education, for, you know, small things like planning their trips. Is it changing the planet yet in a huge impactful way? I wouldn't say so. And that's where the hype is, like, the marketing material is getting a bit too grand.
14:01 Speaker 03 Yeah, it seems like a good productivity tool right now. I mean, when I talk to like software engineers, there's multiple companies now doing AI for software engineers, so like founders at those companies, for example, and even people in the big tech companies that have rolled out their own kind of AI coding tools internally. Engineers are seeing meaningful productivity gains from these copilot-like tools, right? I mean, I feel like that's a pretty good example of the near-term impact that we're seeing. I haven't seen a lot of productivity gains in my own workflow personally, but I hear it from engineers pretty regularly, actually.
14:36 Speaker 04 Yeah, I would say that that's definitely the case. I am old enough, AKA young enough, to remember when GitHub Copilot came out and everyone was up in arms about it using their code as training data. I feel like maybe that was like two or three years ago. And now it feels like everyone uses GitHub Copilot because it's so useful and only getting more useful. And that's what I hear from the AI experts, is that where large language models are really, really good is with coding. Yeah, it's a very practical use case.
15:04 Speaker 03 And another difference with crypto is, you know, a year after the crypto boom, Apple didn't come out and say, hey, we have a cryptocurrency, right? Whereas what we just saw is them fully embrace generative AI at WWDC, just shortly before we record this, and announce a partnership with OpenAI. It'll probably do others to put ChatGPT in Siri. When you have, you know, arguably the most conservative, brand-protective mega-cap tech company in the world embracing this tech, to me, that's also a validator that says, you know, maybe this is here to stay. Maybe your beat is safe, Kylie, is how it makes me feel.
15:44 Speaker 04 Yes, totally. That's exactly, I think, how a lot of us reporting on the tech, especially at The Verge. I've never covered an Apple event and this company is a machine. But I was talking to my editor Jake about this exactly is that it seems like they're the last holdout and now we have AI for Apple, Apple Intelligence. You know, that's, it's much different from crypto in that sense.
16:07 Speaker 03 We need to take a quick break. When we're back, Alex and Kylie break down all the key players in AI today.
16:18 Speaker 01 This episode is brought to you by Subaru. You know that place your mind always wanders to? The lake, the mountains, the homes of the people you love most? The completely redesigned 2026 Subaru Outback is built to help you get there with legendary capabilities, standard symmetrical all-wheel drive, and the confidence to go further. The 2026 Subaru Outback. Love never looked so good. Visit Subaru.com/Outback to learn more.
16:43 Speaker 05 Flowing ad budget on metrics that look great till the CFO sees them? That's bullspend and marketers are calling it out in. Dashboard, confessions.
16:53 Speaker 06 I remember telling my boss, it'll be good for the brand when leads were slow. Yeah, it, it wasn't.
17:01 Speaker 05 Cut the bullspend. 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.
17:15 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 blog, 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 plus.
17:43 Speaker 02 We're back with Verge Deputy Editor Alex Heath and Senior AI Reporter Kylie Robison, making sense of the current AI landscape.
17:50 Speaker 03 Let's get into kind of some of the key players. I think we have to start because it kind of sets the stage for everyone else with OpenAI. And, you know, going back to talking about EAC versus Dcell, I think you would probably put Sam Altman, the CEO, as kind of the poster boy for EAC, right? He is the face of go, go, go. Commercialize this tech as quickly as possible. We've seen the tension that has played out at OpenAI as a result, considering its roots. But there's no question that OpenAI is the reason we're talking about this right now. They've made everybody wake up. Everybody's trying to put transformer technology and LLMs in their products. They still seem like they're ahead.
18:35 Speaker 03 They still seem like they are where the best researchers in the world want to work. For example, it's just something I pay attention to closely, the talent flows. Their models are still the most performant, it seems. They definitely are the most ahead in terms of consumer commercialization, right? I would say the ChatGPT app and experience is the best way to interact with one of these LLMs right now. And they have distribution for a startup. It's amazing. They're on every Microsoft product through that partnership. And now they're going to be on Apple devices later this year. Do you think that that lead that they have, does it stay? Do you see others quickly creeping up and trying to take it?
19:15 Speaker 04 Well, I mean, just the history of tech business, of course. I mean, it always happens. It's inevitable and it's always the ones that you're like, wow, how could HP, how could Intel not be the coolest, sexiest company on the planet forever? Yeah, of course, they're not going to remain the market leader forever. And they're kind of enabling that competition in a way.
19:36 Speaker 03 How are they enabling that exactly?
19:38 Speaker 04 For instance, we see perplexity. The people at perplexity, my opinion here is that they are really, really rabid to compete with OpenAI and be better than them. And I think a lot of people in Silicon Valley, a lot of young AI founders are rabid to be like OpenAI. And I think that's how you get a lot of smart startups. A lot, 99% of them are going to die. But I think a lot of people are rabid to use that technology and make something useful. But OpenAI, of course, is the market leader. They have taken so much power. I am so shocked in only my month of reporting on this, that OpenAI has struck this Apple deal.
20:20 Speaker 04 And I remember during the Scarlett Johansson stuff, which I'm saying I remember, it was like a couple of weeks ago.
20:26 Speaker 03 Again, it's an eternity, right? Every day feels like a year.
20:29 Speaker 04 Exactly. So with the Scarlett Johansson stuff, people were speculating, will they lose this Apple deal? And I was like, no, I don't, I don't actually think so. I think they're going to be the market leader for quite a while. But of course, at some point, they're, you know, someone's going to usurp them. What do you think?
20:45 Speaker 03 I think you're right. I think they run the risk of being Yahoo to the next Google, right? They run the risk of being the first mover and waking everyone else, but then losing because of something like distribution or more backing, you know, something like that. I think they have room to run. I think they're still considered the best place to go to work if you care about commercializing this. I think the drama with OpenAI that is so juicy is that you've got the kind of the factions, the dueling factions, the EAX and the D cells, the Ilya camp, who has been defenestrated and is now no longer there. But you have the roots of OpenAI as this nonprofit, safety-focused research lab.
21:31 Speaker 03 And still legally set up that way, but also now has a CFO that just used to be the CEO of a public company. Microsoft owns basically half of it. There's a lot of tension in not just the people and the personalities, but the literal structure of OpenAI. And I think we're going to see all of that come to more of a head before it gets cleaner. The firing and rehiring of Sam last year was kind of, they call it the blip internally at OpenAI. It was just like the first literal blip, I think, of what is to come. And can they survive that? Can they come out bigger and stronger as a commercial entity? I think remains to be seen. They certainly have every incentive to. You know, we don't like to say a ton,
22:16 Speaker 03 even in coverage, is like, people are just getting fabulously wealthy off this stuff before any company that we cover, really, except for the large mega cap tech stocks are making a profit. And there's just a lot of money riding on, on OpenAI. I think the most interesting dynamic, like, from a technology perspective with them is the rise of open source, which maybe we should go to next, before we get into Google and some of the search stuff. You've got, I think, Meta on the other side, doing this kind of interesting scorched earth approach, which is releasing models for free and commoditizing what used to be OpenAI's kind of core moat, which was cutting edge models that were closed that
22:58 Speaker 03 you had to pay for. And we've seen OpenAI kind of actually change its business model in the last several months, where they put their leading model for, oh, out for free for everyone, which is something they didn't used to do. They used to gate the most advanced models. So open source is putting a lot of pressure on the closed players. And I think that's interesting. I don't know. Do you have a view of open source right now?
23:22 Speaker 04 You know, when I was interviewing in my first two weeks, the CEO of Hugging Face about open source, they had sent me this graph of just showing the gap of capabilities between some open models and some closed models, the capabilities of two kind of closing. And I thought that was really interesting because those closed models are racing fast to stay above and the gap is getting smaller and smaller. So yeah, I think it's interesting and a trend we'll continue to see.
23:49 Speaker 03 It might end up being that the actual AI part of these AI companies is not as important as the consumer products they build around them. I find it fascinating that two former heads of product at Instagram are now the heads of product at OpenAI and Anthropic. One was the co-founder of Instagram, Mike Krieger at Anthropic, the other Kevin Weil was there during kind of their Instagram's formative scaling years where they copied Snapchat and they really became the hyper dominant multi-billion platform it is now. The recent hirings of those two individuals to lead product at the two preeminent AI labs, to me signals we are entering a whole new era where like it's not really as much about the research. Yes, the research matters because that's how you do stay ahead in the
24:33 Speaker 03 long term. But really, it's about building businesses and sticky products and almost like social network like hooks to retain and grow users. And so it's this kind of like, I don't know, you could call it like Wall Street-ification or whatever of AI. But that feels like the moment we're in right now. I don't know if you agree with that, but.
24:55 Speaker 04 Yeah, I think that's a really good point and something I hadn't considered, but it is what's kind of pissing off the factions in OpenAI, right? They're saying that Sam is taking their research and creating products for capitalism, for money, and that's not the goal. What they think is we are a research lab, we are supposed to be creating AGI for humanity, not creating, you know, products, just like we saw with Helen Tonner, who was a board member at OpenAI who helped lead Sam getting fired. She said, we found out about ChatGPT on Twitter. Those are the kinds of things that I think are the friction inside of OpenAI, the Wall Street-ification of a research lab they hold so dear and why I think so many of them on the safety, the super alignment team have left.
25:39 Speaker 03 Right. Well, I think what's amazing about the ChatGPT just coming out on Twitter thing too is like, yes, the board at OpenAI should have probably known about that, but... To be fair, no one at OpenAI knew this was going to happen. I feel like I can say that confidently. There wasn't some grand plan. They felt like they had unlocked something. They put it out as a research project at the end of a keynote. They didn't make any fanfare about it. There were no press releases. They really stumbled into like this new kind of fire almost, it feels like, which is why the beat is also so exciting is like it happened so quickly and caught everyone off guard. And now everyone's like kind of scrambling and arranging the arranging the troops to go into battle. And I guess, speaking of battle, maybe now we should turn to Elon,
26:24 Speaker 03 who you were covering, I guess, in a slightly different capacity before you came, but you're continuing to cover now that you're here at The Verge. He is out there doing XAI, raised a bunch of money, is wanting to build the, what does he call it? Gigafactory for compute. So the largest cluster of chips to train models. And from what I can gather is actually hiring good people and has a compelling pitch, which is like, just come do frontier research. He wants it to be anti-woke, you know, whatever that means. What is Elon's effect on the industry right now?
26:57 Speaker 04 You know, it's funny, a surprising amount of great talent is going over there. I say surprising because it's incredibly difficult to work for Elon, as we all know. You can't even like get married there because the joke is you can't retain a healthy marriage working for Elon. I just had a conversation with someone in Cerebo Valley, and they said, who would you want to run AGI? If you could pick anyone, I was like, oh, my God, I don't know if I want to pick a single person even. And there's, this person is very like safety focused, very like super alignment sort of person. And they said Elon Musk because he had called for a pause, maybe a year ago now, on AI development.
27:38 Speaker 04 And he has this pitch, his public pitch is that he wants AI to be good for humanity and not bad and not take all our jobs and blah, blah, blah. That was surprising. I think it's mostly a marketing pitch. I think he wants money and power and he's not very happy with OpenAI and the way that they've done things and didn't accept him as a leader there. I think that much is known, but. As for his impact, I mean, we saw he just took a bunch of H100s, processing compute power from Tesla to allocate to XAI. He's doing this Gigafactory. I think he's going to move very aggressively. We know him to move aggressively and build quickly. And he's got the talent to do so.
28:20 Speaker 04 So I'm for one surprised, but I do think he has the capabilities to move fast on his large language model. But I've said this before, anti-woke chatbot is not the most compelling use case that I've heard.
28:34 Speaker 03 Yeah, I haven't, you know, really played around with Grok that much. I do actually think that his unique edge is X.com for training data.
28:43 Speaker 04 Yes.
28:43 Speaker 03 It seems like maybe that training data becomes less valuable over time.
28:47 Speaker 04 Yeah.
28:47 Speaker 03 As X slowly implodes in on itself and becomes an echo chamber. But at least it's real time and it's fresh. But yes, I agree with all that. The fact that Elon co-founded OpenAI, there's got to be a deep sense of maybe jealousy at what they achieved after he threw the towel in on it. It's interesting just to think about what could have been, you know? The fact that Anthropic, OpenAI, XAI were all kind of started as one company, right? And they've all split off into these different, these different views of it is interesting. We haven't really talked about Anthropic. To me, they're kind of a question mark. They seem to be getting some early traction on the enterprise side,
29:31 Speaker 03 but they're way behind in terms of consumer awareness and adoption. I think that's why they just hired the co-founder of Instagram to lead product. But Claude is considered good and technically one of the leading LLMs, but it's got nowhere the mindshare of ChatGPT. And they seem like the lone other independent, semi-independent player that could give OpenAI a run for its money if we're not talking about Elon. Is there room for Anthropic to become a meaningful consumer player here? Or do you think OpenAI kind of has it locked up?
30:04 Speaker 04 That's a good question. I want to say yes, but ChatGPT just is getting integrated into Siri. I mean, that's insane. I think about, you know, my grandma trying to ask a question on Siri and able to ask ChatGPT. It's not asking Claude, but Apple has said that they want to integrate more models. We'll see. Yes, I do think Anthropic can take the lead here. They seem purposely unsexy. Like, I feel like they don't want to be exactly the chatter. Maybe that's changing with this product person joining. They are taking up the enterprise side. They're moving very slowly. That's their goal. They're a public benefit corporation. They don't want to be like OpenAI. They want to be more like a research lab is my understanding with consumer products,
30:46 Speaker 04 rather than consumer product focused and like research lab second. I'm wondering what you think.
30:52 Speaker 03 It's like I said, I think just a question mark. I haven't seen a unique angle that they have that gives them an edge. I browser tab all these different chatbots and it's just, it's overwhelming, you know? Even to just try to think about what do I try right now as someone who's like, my job is really to try these things. It's just a lot and I just don't know how much room there is for multiple all-in-one chatbot assistants. And so that puts Anthropic in a weird spot, I think. But, you know, jury's out.
31:24 Speaker 03 We need to take another quick break. When we're back, Alex and Kylie dive into what's going on with Google, Gemini and AI search.
31:35 Speaker 01 This episode is brought to you by Subaru. You know that place your mind always wanders to? The lake, the mountains, the homes of the people you love most? The completely redesigned 2026 Subaru Outback is built to help you get there. With legendary capabilities, standard symmetrical all-wheel drive and the confidence to go further. The 2026 Subaru Outback. Love never looked so good. Visit Subaru.com/Outback to learn more.
32:00 Speaker 05 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.
32:09 Speaker 06 My boss asked for results, so I opened my dashboard for the only positive sounding metric I had: impressions.
32:17 Speaker 05 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.
32:31 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 blog, 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 plus.
33:00 Speaker 02 We're back with Verge Deputy Editor Alex Heath and Senior AI Reporter Kylie Robinson, discussing what role Google is playing in the modern AI boom.
33:08 Speaker 03 Let's turn to where we've been circling since the beginning, Google, which really invented the underlying tech. As everyone listening to this probably knows, the T in GPT, the transformer. And because Google has been a fat and happy trillion-dollar company for a very long time, they let it catch them by surprise. And now they're rushing to try to capture the buzz that OpenAI has shown. We were both at Google I/O. This was, gosh, feels like an eternity ago. It wasn't that long ago, but where they showed the AI overviews that were rolling out to search, to create the AI summaries. Very quickly, that went off the rails. To be expected, you wrote about it.
33:49 Speaker 03 What is Google learning about LLMs and AI as it relates to search?
33:54 Speaker 04 They kind of put the onus on the user. It's like, you guys keep searching up the darnedest things. Stop doing that.
34:03 Speaker 04 We'll fix that, but just don't, don't mess with our product. Just to go back to Apple, I kind of feel like once Apple launches this at scale, we're gonna see similar embarrassing issues. I think it's kind of inevitable. At this huge scale, millions and millions of people are tinkering with your product. Of course, they're gonna Google how to eat rocks and how to get the cheese to stick. And you're gonna get some weird responses. What are they learning? I think that case to me was a bit disappointing because if you're not ready to launch this product at the scale that you're going to launch it, then why? You know, it's for for I O to impress people to, you know.
34:45 Speaker 04 I don't know what they've learned, if I'm being honest. You know, they did clean up the responses. For example, some of the overviews were coming from the onion. That was the rocks thing. And I was surprised you would think that the incredibly smart and incredibly capable multi trillion dollar company would have figured that one out before launching it. So I think they've they've cleaned it up. Hopefully it provides better and more accurate responses in the future. Maybe that's what they've learned.
35:10 Speaker 03 Yeah, I mean, there's this inherent tension with doing L L M, you know, generative AI stuff in search because L L Ms hallucinate. That's kind of inherent to their architecture. It feels like trying to jam a triangle into a square or something like it's just this. This tech is not good for search, maybe. I don't know. The vibe I got overwhelmingly at I O was that Google feels and is seeing in the data that people really prefer these AI summaries and that they're seeing way more engagement. And that's what's driving them to do it. I can see that because it does abstract a lot of the work of kind of surfing through the web, so to speak. And like we mentioned earlier, there are startups like perplexity that are doing this and showing that it's an interesting new way to do search.
35:54 Speaker 03 But, yeah, I mean, Google is the perfect example of innovator's dilemma where they're just this giant company that has a lot to lose and a lot of revenue to protect. AI is maybe a wedge for OpenAI, perplexity, others to try to attack that because Google can't move quite as fast because of things like telling people to put glue on pizza. Then they have to roll it back. Whereas with OpenAI, we don't have that historical assumption, I guess, that what we're getting needs to be accurate, right? Whereas Google has been building that up with search for 20 plus years. So they they have more to lose and I think they feel that. But I think like we keep circling this like I think.
36:38 Speaker 03 A lot of this just comes down to money. It comes down to impressing investors. It comes down to talent, signaling that you're the place that, you know, talent wants to come work. I mean, it was kind of remarkable. I had this interview with Mark Zuckerberg a few months ago. He was really clear that he wants Meta to be seen as an AGI company too, because he wants the best researchers to want to come work on the hardest problems. And that's how you attract the best talent is you just go out there and say, I'm an AGI company. Whether you are or not, I am under the impression increasingly that none of these companies actually know what AGI is. They have no idea when they will hit it or what it actually means. It's become a pure marketing term. You sound like you agree with that.
37:18 Speaker 04 You know, I'm thinking that actually Cohere, their, their marketing thing is we are not building AGI. That is our marketing, you know? We don't want to build AGI. We're not concerned with these, you know, fanatic terms, which is funny. Everyone's got a marketing scheme against it. I think, yeah, I'm getting a little tired of AGI and I don't think it's like the important conversation to be having. I loved reading sci-fi growing up. I read a little Michael Crichton. We're not there yet. I like imagining it, but we're not there yet. And I don't know if you've heard this. I'm sure you certainly have from OpenAI people, but
37:53 Speaker 04 Sam's only goal is like, we need to get to AGI and we are the good guys. So we need to do everything in our power to get there because it should be us that gets it, like gets to AGI. I think it's just a very strange, you know, as again, I joined this beat a month ago. It almost feels like an outsider just picking it apart and going, what are we doing here? Is that, is that really?
38:15 Speaker 03 We're the good guys. We're all racing towards some undescribable, fantastic, potentially world altering end states. And we need to do it because we're the good guys. There's a kind of sounds like a religion.
38:32 Speaker 04 Oh my god.
38:33 Speaker 03 Yeah, the AI beat is rich. It's a rich beat. Maybe we could end kind of looking ahead, not just like a particular company, but kind of where this is going outside of chatbots and AI summaries and search. So there's a, I think, maybe three things that I keep hearing more and more about. It's the concept of agents, robotics and AI seems to be really hot right now. And then generally, we're already seeing a lot of wearables. We're seeing stuff like the humane pen, which was a disaster, but I think an example of new form factors that can arise here. I've got the Meta Ray Bans with AI enabled. It's not great, but it is nice sometimes. And it's
39:17 Speaker 03 a peek, I think, of where this is going. What about that stuff? What about like future hardware meeting AI? Are you excited about?
39:25 Speaker 04 Yeah, that's such a shame you say that about the Meta Ray Bans because I just ordered a pair and I'm really excited to try them because I've seen you post on Instagram. It looks really cool.
39:34 Speaker 03 They're great in terms of like camera glasses and like being able to play a podcast. I just, you know, it comes back to like the problem with these LLMs is that they just make stuff up. And when you're, that is your only way of interpreting the world around you, a thing that is maybe right half the time, you just end up not using it. It's been my experience. Like if you're not going to accurately tell me what the thing is I'm looking at, and I don't know for sure that you're even 90% right, I just don't want to use it. Maybe that's the journalist in me, but like I just want accuracy. So I think if they get the hallucination rate down, it'll become a really cool form factor. But for the AI piece, but I'm fascinated by robotics and wearables meeting AI and what that looks
40:19 Speaker 03 like. I don't know if it means we're going to have like a humanoid, you know, robot assistant in our home anytime soon necessarily. Are you talking to people out there in SF that are working on problems like these? Do you see energy here?
40:32 Speaker 04 You want to hear the weirdest one I heard recently? Because I asked an AI VC, like, what's the weirdest one you've seen recently? And they use AI, I hope I get this right, they use AI to carve a ice sculpture of your pet, cat or dog, and it costs like $6,000 or something like that. And I was like, yeah, you're right. That was the strangest one. It feels like they're just making ice sculptures and calling it AI. But
41:01 Speaker 03 That's maybe not what I'm talking about.
41:03 Speaker 04 No, I know.
41:04 Speaker 03 Maybe we jump forward a few years, but I it is very weird right now, like even like stuff like the rabbit, right? Like it's like, we're going to replace your phone with this wonky little it's cute, but it's like it's we're in a very weird state. And I kind of love that about what AI is doing right now. It's making just people make weird stuff, which like it tech felt pretty boring for a while up until the last couple of years was my experience. Yeah, I don't know. I mean, I guess beyond hardware, the concept of agents and AI that is not just like talking back to you, but doing stuff on your behalf. That feels like maybe more near term. That feels like a lot of companies maybe have that, but they haven't released it. OpenAI has been talking a lot about this. And that
41:49 Speaker 03 just gets us closer to like, our phones doing stuff for us, proactive assistance. I'm excited for that. I think it's going to create all kinds of problems and assistance destroying relationships and all these things. It's going to be a rich goldmine to write about. But yeah, is there any kind of like stuff in the world of agents that you're particularly paying attention to right now?
42:12 Speaker 04 I want to go back to hardware after I answer this, but also I was imagining a world where we have agents that have control of our phones and we're messaging sources. Can you imagine an agent having control of your phone and messaging sources on Signal? I can't.
42:26 Speaker 03 That's scary.
42:29 Speaker 04 AI agents, yes, it is not my favorite term. I just blogged this recently, not my favorite term, but it is exactly where I think in the near term, this is going. We see it just like on a customer service capacity. But, you know, again, the scale issue and is this ready for the scale? Did you see that story about Air Canada, how they had to pay someone? Someone sued them because their chatbot, their AI agent, told them there was this discount and that wasn't true. And this person sued them and they won. So a hallucination caused this company to lose in court and have to spend a bunch of money. So again, not ready. For scale, not accurate, but it does seem to be where this is headed because you can,
43:15 Speaker 04 you know, as much as they like, don't want to admit it, this is going to cost people in customer service their jobs. You're not going to need as many of them. I interviewed Brett Taylor and Clay Bevor, who run Sierra and Brett Taylor is on the board of OpenAI and he used to be on the board of Twitter, used to be the co-founder of Salesforce and they're running Sierra, which is kind of that exact thing. You go to Nike.com and there's like, hey, how can I help you do a return order? That's kind of the AI agents they're doing. And I asked that question as anyone would, what does this mean for people's jobs? Like, oh no, we're just enhancing their jobs. I think that's the way we're heading, whether we like it or not, because it's going to save these companies a few dollars. Hopefully it gets more advanced and more helpful over time.
43:56 Speaker 03 We have to, you know, this is Decoder, so we cannot escape talking about the legal aspects and policy aspects of all this. Maybe this is where we end, you know, on Neelai's behalf. Like, I also had a, I had a meeting with a lawyer who is specializing in AI stuff and he was saying that his firm is staffing up to prepare for the onslaught of liability lawsuits, like the one you just mentioned, where a bunch of normal, traditional, non-tech companies adopt LLMs, they hallucinate, they do things that are very harmful for the company and these companies get sued. And this was a giant law firm that's creating a whole new division specifically for this. And Neelai will have hours and hours of pontification about this
44:41 Speaker 03 when he returns. But for now, I guess that's probably where we should leave it. I feel like we've covered a lot of ground and you know, maybe when we check in with each other in a few months, six months, a year, who knows where we'll be. I feel like every day is an eternity. So maybe AGI will finally be here next time you're on.
44:58 Speaker 04 I really hope not.
45:00 Speaker 03 Me too. Kylie, thanks for joining.
45:03 Speaker 04 Thank you.
45:07 Speaker 02 Thanks again to Alex Heath for guest hosting today's episode and to Kylie Robinson for joining him. If you have thoughts about this episode or anything you'd like to hear more of, you can email us at decoder@theverge.com. We really do read all the emails. Or you can hit me up directly on threads, I'm at reckless1280. We also have a TikTok for as long as that lasts. Check it out, it's at DecoderPod, it's a lot of fun. If you like Decoder, please share it with your friends, subscribe wherever you get your podcasts. And if you really love the show, hit us with a five star review. Decoder is a production of The Verge, part of the Vox Media Podcast Network. Our producers are Kate Cox and Nick Stat, our editor is Callie Wright, our supervising producer is Liam James, the Decoder music is by Breakmaster Cylinder. We'll see you next time.
45:50 Speaker 08 This week, it pays extra to be a MyLowe's Rewards member. Through October 14, members get $10 MyLowe's money when you spend $200 or more on an eligible purchase during MyLowe's Money Days. Also using your MyLowe's Rewards credit card? Even better, that's an additional $20 MyLowe's money. Use it to save on your next Lowe's haul. Exclusions, restrictions, and more terms apply. See Lowe's.com/MyLowe'sMoneyDays for details. MyLowe's money expires 30 days after issuance, subject to change. Credit offer subject to credit approval.
46:19 Speaker 09 Your heart can tell you a lot about your health. Apple Watch Series 12 measures your heart rate every five seconds with the most accurate heart rate sensing in a wearable. So your vitals app, now with heart rate variability, can tell you when something is off. And your readiness score can let you know when to rest and when to push. Hear the story in every heartbeat with Apple Watch Series 12. The features described are for wellness purposes only and not for medical use. iPhone 11 or later required. Based on an Apple conducted study of heart rate accuracy, August 2026. Visit apple.com/AppleWatchSeries12.
Previous Why Tubi CEO Anjali Sud thinks free TV can win again Next Netflix co-CEO Greg Peters on the streamer's shifting culture and where ads, AI, and games fit in Page updated October 12, 2026 · transcript and summary are machine-generated