Searching for the first great AI app
01:35:26 Share

Support for the show comes from Crucible Moments, a podcast from Sequoia Capital. It might be hard to imagine life before Dropbox, DoorDash, YouTube, or any other tech giant. But none of these companies' success was a given, and the podcast Crucible Moments pulls the curtain back on the tumultuous and consequential turning points of today's most influential tech companies, as told by the founders themselves.

Tune into a new season of Crucible Moments now. You can listen at cruciblemoments.com or wherever you get your podcasts. Support for The Verge Cast comes from AT&T.

What's it like to get the new iPhone 16 Pro with AT&T NextUp anytime? It's like when you first light up the grill and think of all the mouthwatering possibilities. Learn how to get the new iPhone 16 Pro with Apple Intelligence on AT&T and the latest iPhone every year with AT&T NextUp anytime. AT&T, connecting changes everything.

Apple Intelligence coming fall 2024 with Siri and device language set to U.S. English. Some features and languages will be coming over the next year. Zero dollar offer may not be available on future iPhones. Next up anytime feature may be discontinued at any time. Subject to change. Additional fees, terms and restrictions apply. See AT&T.com slash iPhone for details. Support for the show comes from ServiceNow, the AI platform for business transformation.

You've heard the big hype around AI, and the truth is AI is only as powerful as the platform it's built into. ServiceNow is the platform that puts AI to work for people across your business, removing friction and frustration for your employees, supercharging productivity for developers, providing intelligent tools for your service agents to make customers happier.

All built into a single platform you can use right now. And that's why the world works with ServiceNow. Visit servicenow.com slash AI for people to learn more. Hello and welcome to VergeCast, the flagship podcast of the agentic era. That's where you just sent us requests and David has to do them. But we say it's AI. Like every other AI company. I want to be clear. That is how they all work.

Tesla's robo taxis, fleet of human operators waiting in the background to pounce. That's just how it works. There's a lot going on this week. I'm your friend, Neil. David Pierce is here. Hello. Waiting to take your commands. David, memorize this list of names and then make new names. What I want to talk about, which is just your bad websites made AI necessary. Yeah. This is what I've come to. Richard Lawler is here. I love AI. I'm a big AI guy now. I've moved on from crypto.

It's all about AI now. Richard, big AI guy. That's what everybody says. I'm saying, Richard, you have not had AI change that car behind you from Mercedes to Ferrari yet. There is no Mercedes. There is only Ferrari. And therefore, it is a Ferrari already. Wherever Hamilton goes, that's a Ferrari. That is also, in many ways, how AI is being defined. Whatever this computer does is AI.

They should get, they should get Lewis Hamilton to be this. I love just remembering three commands to be able to tell my computer to do one simple thing. It's just the best way to use it. There's a lot of like dumb crypto money in F1. Is there a lot of dumb AI money in F1 yet? Not yet. They've been kind of late on that. I don't know what's going on. That's true. Well, a new season is coming and new sponsorship opportunities await.

I will say that Chrome sponsoring the McLaren and then making the cart literally Chrome genius. It's very good. In terms of like just pure, uh,

actual like branding of the thing inside of the actual product, like crushed it. And McLaren wheels are a great idea, especially when they're not Chrome, but they look like Google Chrome. Yeah. No, but they added actual Chrome to the car. That was this year's big innovation. They put literal Chrome on the car, which is not what you think about when you think of an F1 car. Like what if this looked like a 1950s cattle? It's very good. That's all I'm saying. All right. There's a lot of news, actually I news this week, not,

just Chrome in all senses of the word. And a lot of it is Google. And David, you wrote a bunch of these little bits and bobs. Gemini 2.0 is out. Google is making some mixed reality noises. There's an Android mixed reality thing yet again. Be honest about that. They've tried this before. They're back again. There's a new Jules AI for code. What's all going on here, David? Okay, so...

let's let's break all that into three pieces uh and we should talk about all of them and i think the three pieces are uh gemini 2.0 there's astra and mariner which are kind of like overlapping with gemini 2.0 but are are like products and they're interesting we should talk about them and then we should talk about the xr stuff at the end uh it's

It's called Android XR. I have many thoughts about the decision to use XR, but we should talk about this. But let's start with Gemini. So the big news of this week was Google is starting to roll out Gemini 2.0, which is the successor to 1.5, which I think it first launched in February. So

So like nine months of development later, this is the new thing. And what Demis Hassabis told me, who he runs DeepMind at Google and is in charge of all their AI stuff. And by the way, I talked to him earlier this week.

While he was in Sweden accepting a Nobel Prize, he called me from his hotel room before the Nobel Gala, which is like the worst power dynamic of an interview I've ever had in my entire life. It was really something. I was like, I don't know how to tell you you're wrong on any of this because you're literally, you're going to get a Nobel Prize tonight. But anyway, what he told me is that basically what Gemini 2.0 Flash is is

is roughly performance equivalent to what Gemini 1.5 Pro is. So the way to think about it is it's like a full step up in terms of efficiency and speed and latency, which is stuff that really, really matters with still the same amount of performance as the Pro level tier from last time. Just to unpack some Google product names here, Flash is the little model that's supposed to run...

on phones. That's right. So they, they call it the workhorse model. Uh, it's the one most people encounter most of the time, basically. But so that's not the phones one. That's just the sort of like, here's the one you get flash. There's one below flash that I can't remember the name of, but it is, it is like, it's the, it's the main one. Like when you use Gemini now, you're using 1.5 flash. Okay. Certainly. Uh, nano is the one on the phone. Okay. I just want to be Google product names. Also, I just want to point out they named it flash.

Yeah. Like in a history of technology names to reanimate, they picked flash. Yeah. It's all you didn't like new grounds. look, home star runner was great. Right. And if these, if these AI systems can deliver me home star runner, that would be amazing. If Google made home star runner, the official mascot of all of its AI. And like, that was the voice and personality I was talking to. I would be all in on AI, honestly, like done lock it down. Get these woke AIs out of here. Put strong bed in let's make this happen.

All right. Keep going. Exactly. Uh, but so, so yeah, so it's, it's kind of a magnitude leap in power for them. It's also, uh, like natively doing a bunch of things that were separate in products before. So Gemini 2.0 can now do images natively, which used to be a thing Google did with a separate model. It can do audio natively, which is a thing it was doing in a separate model. So there's just a lot of this like multimodal stuff coming into Gemini. Uh,

which is a big deal for Google in a bunch of ways. But the main thing is like, this is going to be the model that Google tries to put everywhere. It's the one that they're going to try to sell to cloud customers who want to do all this kind of stuff. It's the one they're going to put into AI overviews and search. It's the one they're going to put into Gmail. It's the one that's going to be in Gemini. Like,

Like increasingly Google is trying to do kind of the opposite of what OpenAI is doing. OpenAI is like, we have a bunch of models for a bunch of different things and you can use them all and they all kind of interoperate, but it is like different models for different needs. And Google is just like Gemini. And that is very much like how this company wants to approach AI. And I get the real sense they see this as a big turn in that direction. Wait, let me ask just a foundational question here. Yeah.

There's a big conversation in the AI industry about scaling laws and whether training the next model will result in more capability. You wrote very disparagingly about OpenAI trying to redefine AGI down. I think you actually said we should just make fun of Sam Allman on this show a couple weeks ago. I did. I still believe that. Right. So we're trying to drag the word AGI down to the ground and just be like, anything. This F1 car is AGI. Sure.

Then there's Google and it's their newest model is the same capability, but more efficient.

which seems important, but it's kind of lost in the AGI conversation. Because all of that is like, can we make AGI out of a bunch of NVIDIA chips that we have today? If we throw more data at them, will they get smarter? And Google's like, it's the same smart, but more efficient. Yeah. So I actually talked a bunch about this with Demis, and I'm very glad you brought this up because I've been thinking a lot about this in the context of these announcements. And I think Google's stance seems to be

uh, that there is more headroom to be had. Uh, he, he kind of allowed that it's slowing down that the idea of this stuff, just like linearly getting better with every new model is probably not true, but he did say, he's like, there are still performance gains to be had. There's new stuff we can learn how to do. And he also said that actually what we need is, is the new technique. We don't need more. We need new is kind of what he kept saying that like,

the way that transformers change the way we think about this stuff. Like we need another one of those if we're going to unlock like the real next step change in AI. And he seems to think those are out there and exist and they're working on them. And he has some theories. He didn't want to tell me about them, but like there is room upwards, but it is not just,

just bigger models. What's next? And I think what we're seeing right now, both with what Google is doing and with what OpenAI is doing with this 12 days of shipments thing is everybody is trying to figure out now how to make these things products. Like, how do I get these things in front of you in a way that is useful and valuable and worth paying for or at least using and in some like very meaningful way pays off as a business for these companies that are just pouring money down the toilet?

making these things work. So for Google, one thing I liked about the way DemoSource streaming it is like Google works on all this stuff at sort of a technological, infrastructural level. And then Google

is Google's biggest customer, right? So there's like, he was like, we think of this as, and this is like, you're making a face. This is what everybody says about their stuff. If you're listening to this in your car, just imagine me rolling my eyes so hard, your car crashed. Everyone ever on Decoder has said that about their companies. But like, it's true in a lot of cases. And I think it's true in this case too, that like what Google needs is for decoders

Gemini and Gmail to not run Google out of business because it's so expensive to operate. Right. And so, so the idea of making this stuff faster and more efficient and lower latency becomes really meaningful, not only because Google is in this like incredible arms race with every other cloud company to offer AI services and this stuff is being really quickly commoditized. So being able to say we are the cheapest and the fastest is becoming very powerful very quickly.

But also, it matters to Google's own products in a pretty big way. So he kept getting really excited about efficiency. And I realized he's like, oh no, you're serious. This means you can build this into products in a way that works at scale without just absolutely hemorrhaging cash the whole way. Yeah, but we should talk about those products, and we will. But I just want to say this very clearly. And I'm the person who interviews all the CEOs about their costs. I don't care.

Right. Like we made it cheaper to hallucinate is like a, I mean, honestly, there was a time in my life where that would have been a very compelling pitch, but like, they're not better. Right. They're just cheaper to run. And I feel like I'm just stuck there. Like, that's a big deal. I'm not discounting. It's a, it's not a big deal to put the thing more places and see what kind of product you can build. And maybe all anybody wants is to hallucinate some emails look great, but they're not more capable yet.

Like the big innovation here is that it got cheaper. If Google wants to save some money on AI, they could just turn off the AI overviews on my search results. Like I don't know how many pennies that would save them, but it's got to be at least five cents considering the amount of searches I do. It was really funny. Google, I think it was Sundar Pichai wrote a blog post with all these announcements that was like,

you know, there are now more than a billion people are experiencing AI overviews. And I was like, that's such a different way of putting it than more than a billion people really are excited to use AI overviews. We have foisted this upon a billion people. Yeah, exactly. It's like, this is one of the fastest growing search enhancements ever. And I just kept thinking about you two being like, this is our most successful album ever because we shoved it onto everyone's iPhones. It's like...

I don't know that that's the flex you think it is, guys. So we should talk about the product. That's the thing that really leapt out to me is we have very quickly moved on from talking about the models getting more capable and we're worried that these are going to be super intelligence level extinction events for us. Maybe Sam Altman will just say it's AGI because he wants to because that'll get him out of a Microsoft contract or something will happen and it'll boost his stock price when they go public.

to, we've made it cheaper. We can put it in more products and hopefully we can all start making some money. And that that's fine. Like, great. Like very cool. The products still aren't there. They're not very compelling yet. Maybe some, someone just has to figure it out. And then it, it just feels like I keep making the comparison to Bluetooth. It's like, they keep telling me Bluetooth is awesome. And I'm like, but these headphones are bad and we're just stuck in that loop. So tell me, I guess, tell me about the headphones or the headphones. Any good.

Well, okay, let me describe the headphones to you. Google launched four new pairs of headphones. One is Project Astra, which we've seen before. We've talked about it a bunch on the show. It's basically like the most sort of all-encompassing, ambitious version of Google's AI idea. It's visual. It's listening. It has memory. It's the thing you like...

walk around your house with and then you're like where did I leave my glasses and it tells you because the camera's higher glasses like that's the idea they're able to do some new stuff on there it's also now connected to things like Google Maps which is cool so there's like they're starting to be able to plug some of these things in and that's where like again having Gemini underneath all of this stuff makes that

doable in a way that's been much harder in the past. So that's one. Thing number two is what's called Project Mariner. And this is, it's a Chrome extension. Google calls it an experiment. They have a million different terms that all mean some version of prototype. We can kill this without you yelling at us. Right. This is like the most prototype-y prototype, but it is like an agent in the way that we've been talking about agents for a while in that it can like go and browse the web and do things for you.

Kylie Robison on our team got some demos of this and her take was basically it works. It's really slow. It's kind of wonky. I'm not sure this is meaningfully better than me just like looking at some web pages myself and

But one of the demos they give is like, look at these web pages and find the contact emails for me. And it will actually go and literally click around the pages and try to find the email for you. Can we talk about that one for a second? That was the one I was talking about right at the beginning. Yeah. The request. And you've got to see the video and we put a screenshot of it on the site. But the request is absurd to me. It starts off with memorize this list. What does memorize mean to a bot? Yeah.

If I gave you a list, I assume that you remember it now, if you were a computer, because that's what you do. No, no, no. These computers are on drugs, Richard.

And now I have to tell the computer to memorize. And then somewhere else in the command, it also said, remember again. So now I have to remind you to remember this thing, to go do a thing. And then in the context box, it tells you the results are unreliable, like right below it. So I'm telling an unreliable party instructions that it can't follow. If I was going to have someone screw up a task, I would just do it myself. Like what is AI doing for me? But no, this lets you screw up higher value tasks. Yeah.

Right. This fails to get a bunch of email addresses while you like, you know, crash a bike like you're doing something else. There was a good one on there, too, in the in the Project Astra demo video. It starts and ends with a guy trying to get into an apartment building. And the first thing he does is he's like, look in my email and find the the code to this front door and then remember it.

And it's like, I had the same reaction. It's like, first of all, it's just in my email. Like, it can just look in my email every time. Who is remembering anything? And then at the end, he's like, what was that code? And it's like, the code you asked me to remember. And it's like, no, it's still just the code in my email. Like, what are we accomplishing? You just look at the email again if you want to. It's fine. Like, now I have to remember the conversation I'm having with the computer about the thing I'm trying to do. Right. I have to tell you what I've told you so that you can tell it to me again. And it's just like, what is...

Just tell me the code. Just what is the code? We fixed this. You let me pin notes on my phone screen and then I had the code. Yeah. There's a lot of weird UI around this stuff. And the question of like, what is this actually accomplishing that is AI and is new? Hard to say. Again, I can't stop talking about the fact that

hacking your way into like pretending to use a web browser on my behalf is not impressive and not interesting. Like do other things. Well, so let's split these two up. Astro is interesting, right? Cause we're the same day that Google announced all this stuff. Apple announced iOS 18.2 released iOS 18.2 with visual intelligence and the, Hey, look at something and tell me about it. That's the world we live in now. This is the new great AI feature, which, you know, other companies have had before.

But now it's built right into the iPhone. And then Google is talking about Astro with it. It seems like Astro is very much the future of Google Lens, right? You're just like looking at stuff. It's talking to you about it. You're having a great time. But it is very much like I looked at a picture. I scanned all the text in the picture. I did some visual recognition of what I'm looking at. And here's some information. You can have a conversation with me about it. That, to whatever extent that is super useful for people, I get it.

I think there's still just what I keep calling the U S capital problem. Like you ask Apple visual intelligence, what happened at the Capitol building on January 6th? Like that answer is pretty dicey. Right. And like, there's a universe of actors who would love to change what that answer is across the political spectrum. And none of these companies have contended with it.

Astro's got the same problem. Everyone else has the same problem. But they're headed that way. And Google Lens has been headed that way for a long time. Wait, do you think it's like a real mainstream use case that a lot of people are going to walk by the Capitol and be like, what happened here on January 6th? I think it's a real mainstream use case that you're going to be wearing a pair of glasses in the future that purports to augment reality and everyone lives in a weird custom political reality. Yeah, sure. Whether or not you're looking at the Capitol or not. Granted.

Right. I mean, I'm not going to go into it because it's super dicey and loaded and we should probably just do an entire episode on it someday. But you can just think of the infinite questions that come about when you look at something that are ultra loaded and that someone will start a culture war over. Right. Yeah. Every in your house, you can just look at stuff and be like.

You know, like, are the Cowboys good? There's just like a lot of questions underneath that that are like really hard. Like, Hey, is that Aaron Rogers? Tell me about his ideas. Like you just go for it. Like just normal everyday things. Um, and that to me is, it's great that we're building technology for an Astra is just very much the future of Google lens, but there's this whole universe of those problems that come with, I'm looking at stuff. Tell me about it.

Who will augment reality is like a great question. Yeah, I think it's... I think there are sort of two separate pieces of this kind of thing. And I'm really bullish on one and kind of bearish on the other. And I think the like...

how to logistically get through your day thing, I think things like Astra are going to be really useful for, right? Like there's a moment in the demo video they just released where the guy holds up his phone and he points it at a bus that's just going by. And he's like, well, this bus kept me to Chinatown. There are a lot of like complicated problems to solve technologically there, but that's a really interesting UI problem because that's a hard question to answer. And if the AI assistant can answer it, that's very cool. And the like,

stuff around me and how I get places and like, just like how to do your day. This kind of thing can be really useful for it. If it works. I mean, can I just hold on the bus example for one second? One, ideally the, will this bus get to me to where I'm going? It's not a culture war problem, right? It's a question with an answer. Like, will this go there? And it's an answer that Google already has in a million structured databases with deterministic systems running them. Sure. So you can just ask Google maps, this Google, uh,

has all kinds of public transit infrastructure projects that have put sensors on buses and there's open systems and you can build products and all that. And they've just built that stuff to enable maps a decade ago. And so now you've got a conversational layer that is a new interface for that, where you're just like, look at this, I'm asking about it, figure out what that is. And then it parses that into a structured query for its existing Google map systems and delivers you the answer in a natural language way. And so really what you've just

built a new interface for the existing good system that works. And I think that's very power. All the good AI stuff is exactly that. Right. And I just, that's like one. And when people talk about it, it's being a platform shift. Like that's the one I see. Like we made a click wheel and now we have iPods. We made multi-touch and now we have smartphones. We made natural language UI that is pretty good. And now we're going to get a bunch of other stuff. We made a UI that can listen to a song and tell you about it.

Right. There's, there's like a lot of that stuff that I think is cool. It's the next turn where it's like, because it's a conversational natural language UI, you expect it to have conversations with you and like be a useful companion. And that's where it just always seems to fall down. Yes. Because the systems aren't actually intelligent. And that like, that's to me is like it, it, we keep sliding from one to the other. Like here's a kick-ass way to use Google maps into, this is your best friend that will, you know, if you ask it nicely, we'll try to have sex with you.

You're just like, well, that's a pretty long road, actually. We assume because one, thus inevitably the other. Yeah. And I think everyone needs to stop doing that. And I think one thing that Apple is doing right to its credit is it is stuck on the idea that these are features that should be useful and not best friend companions that you have. And I think...

for the foreseeable future that is you should a pick one of the two and not try to do both and i think the one you should pick is like try to be helpful because we just keep seeing stories about the friends being weird and problematic we just ran one of those stories from josh as like a long yeah like it's it's a mess and it's that stuff is real but like but in this case like again i think

like down every road lies a culture war. I agree. Right. Like if you, if you, if you follow it long enough, there's a culture war there somewhere, but like there are things along the way that work and are useful. And I think Google is starting to poke at some of that stuff in pretty, but I'm like the culture war is like not even, I'm not even saying it's like politically dicey. I'm saying you're somewhere and you're listening to that new choral version of like a prayer by Madonna. And you're like, is this song appropriate for church? And some people think the answer is yes.

Because they're performing it in churches in America today. And it's like, well, you just got to, you got to just, the robots got to unpack that box for you. Right? Like, and I don't think these companies are ready for that, which is what will truly make them useful is by giving an answer you might not like. And I, they're super not ready for that in the current political climate. And that's like,

that's when you make the turn. Like that's when it actually becomes a useful assistant to you. Go find me some information, even if I don't like it. As a slightly lower stakes version of that question, one of the demos- I'm saying is Like a Prayer a song about doing it is like not a high stakes problem. Depends on where you are. I suppose. One of the examples that they showed with Project Mariner was someone pulling up a recipe in Google Docs and saying, add the veggies from this recipe to my Safeway cart.

What if one of those things is a fruit and I forgot that it was a fruit, not a vegetable, or I just disagree. Yeah.

Now that's a high stakes. Yeah. Add these sandwiches to my cart. One of them is a hot dog and the thing just like a data center explodes somewhere in America. This is something that will happen to someone. Yeah. So that's, I just want to like call that out. Like the, there's the thing we want them to do, which is not just a technology problem, like a straight up culture problem, like a society level problem. And then there's like the really interesting part, which is build a better interface for systems that already work.

And then there's this weird middle ground, which I think Mariner is, which, you know, Copilot is at Microsoft, all these agentic systems, which is what if we just use the internet for you? And that's how we solve the problem of we need something in the middle that works. And David, I think this is where you're getting to like, what a stunning indictment of web design.

Because you're like, well, the Internet's pretty broken. So it would be better if some unreliable hallucinating robot used it for me. Right. The assumption is that you can't find anything, which is a very funny thing for the company that made Google search to say. But yeah, we're at a moment where it is perceived to be easier to ask this thing to go find a contact email address than it is to just go find the contact email address.

And again, like I think a place we are landing that is very funny is that all these companies are slowly discovering that the most useful thing their tools do are really boring and really like back of house accounting firm kind of stuff. Uh,

And that turns out to not be very interesting to like a mainstream public that you need to get excited about using your products. So they're stuck in this place of all these companies are actually playing like a thoroughly B2B game and just pretending that they're making like huge strides toward mainstream consumer products. That's all fine and good. And I think like, Nila, I think you're really like, you're right to be hung up on the fact that these things get some things wrong, but also like,

people use them anyway. Like this thing will write my email for me and it will make two weird mistakes, but then I can fix them and send it. And that's better than having to write the email from scratch. Or not fix them. As we are learning every single day, the answer is not fix them. The answer to some extent is read them out loud on cable news. The answer is Hunter DeButts will save us. But yeah, and like,

Hunter Debuts will go down in history as like-- - I refuse to explain. It's there, you can look it up. - Yeah. Ironically, if you Google Hunter Debuts, you'll get correct information about what we're talking about. But anyway.

I just think we're at a moment now where the question for all of this AI stuff, just to bring it all the way back around, is what is any of this actually useful for? And I think the truth is the models that we have now are better than we need for a lot of things and vastly, like a million miles away from being good enough for what everybody keeps talking about. Right? So now the question is, okay, we're not going to build the perfect thing for a long time, if ever. Right?

What can we do with what we have now? And that's where things like efficiency come in, because at some point, like running a business costs money and you have to be able to do these things. And if you can make them cheaper, more people can use them and find new things to do and like on and on and on. But the question is like, OK, if you just take the technology we have right now and you don't pretend it's God, what is it good for? And that is like that is the question Google is starting to try to answer. It's the question OpenAI is starting to try to answer. That is like the question of 2025 is like, what is any of this actually useful for?

Because there are answers. They're just not as like, they're not God and they're not fire and they're not the industrial revolution. So what are they? It's like a tour guide and an extremely unreliable personal assistant. Which to be fair is not nothing. It's not nothing. And I can see why they're compelling and I can see why. I mean, just the demo. I mean, Richard, you already brought it up. Like the Google demo of Mariner where you're just like looking at a recipe in Google Docs.

and you just ask it to go build you a shopping cart on a grocery site and it just does it for you. Neat. Legitimately neat. Very slow. So slow that the Google PM literally saw Kylie noticing how slow it was and said, that's the elephant in the room. That it's ponderously slow. Right? They know. They know it. They're not hiding from it. It's a research demo. But I'm stuck at...

The systems are not themselves capable. So they are free riding on other systems that can do the thing. The grocery site has to exist and work. And it needs to have a database of all the food in the grocery store. And then you need to be able to put it in a cart and like it, like all that has to work and be profitable and sustainable. And then on top of it, you're just like, and then this robot will cost $20 a month and like do it for you. And eventually you're just like, what if,

What if this undercuts all of that? I keep calling this a DoorDash problem where DoorDash has to exist for a bunch of these services to get you a sandwich. And if you just sort of delete DoorDash's business by cutting them out of the equation and taking you, the customer, away from DoorDash, like, how is anyone going to get a sandwich? Like, you need this big, traditional, deterministic, logical computer to

With a database that is reliable doing a bunch of work. So your natural language interface can be useful. And I don't think anybody has solved the problem of like, how do you keep DoorDash in business? Including DoorDash. I'm not just picking DoorDash. Whatever. Uber, right? Like, Rabbit's just like, we're going to click around Uber's website. All of these middle companies, these like Web 2.0 companies that digitize the interfaces for very physical things. Like,

They were pretty vulture-y in their times, right? Like Uber is the biggest taxi company in the world, doesn't own a single taxi. Airbnb doesn't own a single hotel. Like that's the joke about those businesses. And now they're under pressure from AI systems that might use their interfaces and never show them to a user. And it's like where... This has to go somewhere that is actually useful for people. Well, it's a weird thing because you have...

All of the AI companies would agree with you, but they see that as a strength and not a weakness, right? What they're saying is, okay,

We are going to have an army of service providers who just invisibly do the job on your behalf. And you don't have to worry about who they are. You don't have to go to the DoorDash website every time you want to order McDonald's. You just order McDonald's and we'll get it done for you. And DoorDash or whoever will be our provider of choice to do that. They would call that a victory, right? Because I, as a consumer, I shouldn't have to interact with DoorDash to get McDonald's. I should interact with McDonald's.

But what they've actually done is, A, insert another step, and B, completely commoditize all of these providers in such a way that they will eventually just crash them all in an incredible race to the bottom to be one of these providers. And so it's like when all the carriers got really nervous about being dumb pipes, and so they bought content businesses, it's like there is another turn of that coming where

Uber is going to be terrified of being abstracted away by all of these AI companies. And so it's going to do more to try and get you to use Uber. And it's just going to get really weird as a result. Or it's going to just raise the price. Or it's going to raise the price. Or it'll raise the price for the AI systems. So an Uber costs more if you book it through AI.

Than if you book it yourself in the app and look at an ad for Uber Eats. Yeah, it's just the next generation version of the App Store tax. And I think we're just going to see a bunch of this stuff play out. And every time everyone talks to me about these agents, and Google in particular is in a very privileged position because they run Chrome.

And they run search and like one of their, their deep research preview that they released in the same set of releases where it's like, go research a topic for me. And it comes with a research plan and then it goes searches the web for interesting results and then delivers that research back to you. That's just perplexity.

Right. It's just perplexity by another name. And perplexity is in a bunch of trouble because it's not paying any of the sources it's using to scour the web in that way. And it's like, at some point, all of the people who make websites, us included, and I feel like disclosure, the company has some sort of meandering deal with open AI that hasn't really resulted in as far as I could tell. But like, there it is. Like all of these companies are going to say, well, we make the information.

We make the interface to book a taxi or whatever it is. You have to pay us because if you just take the customers away from us, we will go away and the bots will be useless. And I think that just that economic reality will result in some payments, whether or not the lawsuits or whatever results in payments or blocking robots at TXT. But it's come. It's it has to be coming.

And I think the, I think it's going to like follow the same pattern. Everything else follows like Uber will be exclusive to copilot and door dash will be exclusive to Gemini. And like, it's that part's going to suck for a long time. Yes. Or we're going to, somebody is going to figure out how to like do the hack to make it work.

whether the companies want it to or not. And then we're going to have an entirely different kind of mess. And then a new generation of college students will be radicalized by Napster. And then that's how we will be for Vendetta, my replacement. It'll be great. We got to take a quick break, but talk about this XR thing really quickly. Yeah. Okay. So, you know, Android, how it's Android. Imagine if smart glasses were a thing. And if Google just did Android again for those, that's what Android XR is. And basically, Google has been at this smart glasses thing for like

15 years now, maybe, has tried a bunch of things. Do you remember Daydream? Daydream is one I had forgotten about that I just randomly rediscovered today. Daydream existed. I wrote a whole thing about it for Wired a million years ago. I completely forgot about that. But anyway, so Google announced like a developer preview of Android XR. It's basically what you think. It's immersive versions of some Google apps and some Gemini stuff, basically. The underpinning of all of this

is the assistant Gemini stuff that you might want to wear on your face. That's sort of the big theory. They have a bunch of hardware partners. I've been hearing from like every company that wants to sell you glasses over the last two weeks being like, we're doing a thing with Google that we can't quite tell you about yet. That's Android XR. So starting next year, I think we're going to start to see real development and hardware products. It sounds like Samsung is going to be the first out of the gate with like a real honest to God Android pair of glasses. But

It seems very clear that like Apple is trying to be the Apple of your face with the vision pro and Google is very happy being the Android of your face. I think the question of whether Google is going to make hardware is really interesting. Google's made a bunch of prototypes and gave us some indication this week that it is interested in them being more than prototypes. And if you want to do all this Gemini stuff and project Astra, like you got to have glasses. Like if, if this AI future is,

that everybody's imagining and trying to sell you on is going to come true it's going to be glasses it just is i i'm a hundred percent convinced i'm just saying it's a decade from now there's an eighth grader one of our children will be in eighth grade at some point uh they're on their eighth grade trip to dc they're wearing their glasses they look at the capitol and say what happened here what's what are some notable events in american history does it say an insurrection or does it say it was a day of love and are the answers on two different eighth graders different

And these companies are racing towards a future where they are going to have to answer that question, regardless of whether they make the hardware or not. Aren't you just describing the internet, though? Like, everything already works like that. Right, but like... And like, is it ruining society? Sure. But like, I don't know that smart glasses are a new problem here. But you're not literally experiencing... Maybe that is more nihilistic than I want it. Like, I was going to say not everyone's experiencing a bespoke reality. And then I just thought of a generation of Americans who super are. So...

Maybe, but I'm just, you can build, want to build this stuff. And then underneath it is this very deep question of who gets to augment reality. And do we all experience the same one? And they're just not ready for it. Fully not ready for it. Especially when the answer is different. If you have Android on Lenovo pair of glasses versus Android, Samsung pair of glasses and the Samsung pair of glasses is like, there's a second reality here that we made. It sits right next to Google's reality. And it has Bixby in it. Like,

It's slightly brighter. We got to take a break. We'll be right back. That's yeah. I'm just going to, I need to walk around the house a little bit. We'll be right back. Support for the Verge cast comes from the Washington Post. With so much happening in the world, it can be challenging to distinguish between what's real and what's not. Having a trustworthy source to help you separate fact from fiction is essential. That's why you might want to check out the Washington Post.

Right now, you can go to WashingtonPost.com slash Verge to subscribe for just 50 cents per week for your first year. The Washington Post offers substantial and considered coverage on a wide array of topics. The paper can be a great place to find thoughtful stories about Capitol Hill, the economy, climate change, and so much more. I'm a subscriber to the Washington Post, and my favorite thing about it is their clean homepage. I can quickly get caught up on the news every time.

Now is the time to sign up for The Washington Post. Go to WashingtonPost.com slash Verge to subscribe for just 50 cents per week for your first year. That's 80% off their typical offer, so it's truly a steal. Once again, that's WashingtonPost.com backslash Verge to subscribe for just 50 cents per week for your first year. Support for the show comes from Crucible Moments, a podcast from Sequoia Capital. We

We talk a lot on this show about tech companies that shape our world, for better or worse, but the stories of how these companies were built are often surprising and just as interesting. Hosted by Rulaf Bota of Sequoia Capital, Crucible Moments pulls back the curtain on some of the most tumultuous and defining milestones in tech history. You can go inside the decision-making and hear the unvarnished histories of some of tech's most influential companies told by the founders themselves.

Did you know YouTube started as a dating site? Hear how even after pivoting, the founders had trouble getting anyone to upload videos at first. Or how two electrical engineers decided to take on genetics and developed Nutera's breakthrough prenatal test. How Robinhood had to reinvent itself in a market downturn. Tune into a new season of Crucible Moments now. You can listen at cruciblemoments.com or wherever you listen to podcasts. Support for The Verge Cast comes from AT&T.

What does it feel like to get the new iPhone 16 Pro with AT&T next up anytime? It's like when you first pick up those tongs and you're now the one running the grill. It's indescribable, like something you've never felt before. All the mouthwatering anticipation of new possibilities, whether that's making a perfect cheeseburger or treating your family to a grilled baked potato, which you know will forever change the way they look at potatoes.

With AT&T NextUp Anytime, you can feel this way again and again. Learn how to get the new iPhone 16 Pro with Apple Intelligence on AT&T and the latest iPhone every year with AT&T NextUp Anytime. AT&T, connecting changes everything.

Apple intelligence coming fall 2024 with Siri and device language set to U.S. English. Some features and languages will be coming over the next year. Zero dollar offer may not be available on future iPhones. Next up anytime feature may be discontinued at any time. Subject to change. Additional fees, terms and restrictions apply. See AT&T.com slash iPhone for details.

- Support for the Verge Cast comes from the Life360 Tile Tracker. There's a lot going on during the holidays already, so wouldn't it be great to not have to worry about the safety of your personal items? Well, Life360's Tile Trackers can help you eliminate some of that stress by making sure you know where your items are at all times. That way, you can put all your energy towards deep breaths as you navigate the family trip to visit your in-laws. If anything gets lost or stolen, just ring your tile and you can track down its location on the map. That's peace of mind all season long.

The Tile Tracker is the first and only tracker with an SOS button, which means you can triple-press the button in the center of the Tile device to discreetly send an SOS alert to everyone in your Life360 circle. And unlike competitors, Tile won't alert thieves to its presence via notifications. So take the stress out of your holidays and keep your family's most prized possessions safe. Family-proof your family with Life360's Tile Trackers.

Visit Tile.com today and use code VERGE to get 15% off. That's Tile.com, code VERGE. All right, we're back. My feeling is that Bixby would say it's a day of love. No further commentary. We should really do like a political alignment matrix of all of the different virtual assistants.

I feel like if I sat down and thought about it for five minutes, I would end up with really strong opinions. Who did Tay vote for? Yep. So my question is, do you want to go viral on X in this specific way? You're like, we're going to milk the last drop of traffic from this one dead platform. And it's up to political alignment chart of different AIs. No, thank you.

All right. There's more AI news. It's just a lot of things got released this week. OpenAI released Sora. iOS 18.2 is out with chat, GBD integration, visual intelligence. What do you want to start with? Let's do iOS 18.2 first. Okay. I mean, it's very familiar. I think we had a moment this morning where like, how much should we cover this? Because people have seen it. They've used it. It's been in beta. The release candidate was last week.

But it's out now. If you have an iPhone 16 or up or even iPhone 15 Pro, you get visual intelligence, you get Genmoji, chat GPT integration is there. Joanna Stern wrote a very funny column where she's like, you invoke chat GPT by saying secret words. If you ask Siri one way, you get a regular Siri answer. But if you ask Siri, write me a list, chat GPT is the list for you, which is very funny. I've been using this stuff, David, I think you've been using this stuff

What do you think? Richard famously uses Windows Phone, so he's got only knows. It's the only way to live. I will say two things. I think one, the single coolest thing about iOS 18.2 is the ability to change default apps for various things. Like when you tap on a phone number, you can change which app it goes to. Everyone should go examine all of those defaults because if you get that kind of thing right, it makes your phone a lot better. This is the sort of thing Android has been better at than iOS for forever. And thank you to

occasionally confuse EU regulators for making this happen. The chat GPT thing I think is interesting. What I've found is that when I ask compound questions, it punts to chat GPT more often than not, right? Like if my question kind of has two parts or has a thing and then a modifier, like one I just did the other day, it was like, if I ask for a chocolate chip cookie recipe,

Siri will just kind of pull it. And I'm going to not say that word again because it's just activated three devices. But if I ask for a chocolate chip cookie recipe that maybe will taste like I haven't tried before, then it punts to ChatGPT. And then it's a relatively consistent experience. Actually, one thing that I've liked is that when it goes to ChatGPT, the thing that it brings down still looks like

the normal Siri response. So I think in that, like the, my mental construct of what's going on is very confused in a way that I don't like, but the actual user experience of it, I think is actually pretty good. Yeah. I mean, in terms of like, this is a good way to get better answers from a Siri response.

And so you just get, I think that works to some extent. Yeah. Like, do I have feelings about the idea that my stuff is going to chat GPT over and over and over again? Sure. But if you don't, that friction should go away and there isn't actually as much friction as I expected. So I think that's probably on app on balance. A good thing. Yeah. I mean, again, I've had the action button on my phone mapped to the chat GPT voice assistant for ever. This is just like a slightly faster version of that for me.

with like more annoying animation. Like I actually think the new Siri animation is so overwrought. I agree. I missed the little swirling ball. The thing where it like it now takes over my whole phone screen. I don't like it. Yeah, it's very much like jazz hands. It's like get ready for an experience. It's like I don't need this to happen. But it just means that it's going to take a little bit longer to load. Right, and then you talk and then it does the thing where it like fills out the words for you.

There's some amount of obfuscating how slow it was with the swirly ball made it appear fast. You know how progress bars are mostly designed to go really fast for the first part and then they can be slow at the end, but you've already perceived it being fast? That's very much what the old Siri was doing. And now it's just like, everything's glowing and you can see the world. And it's like, I don't need to see the mouse running on the wheel, guys. Just let me know when there's an answer. I think that's right. I had actually not

really ever thought about Siri as a spinning Mac-style beach ball of loading things until it switched to this. And then it is so desperate to make it seem like something is happening at all times because it's slow. So it's just like, look at these beautiful pink and blue swirlies

until it finishes, which it will do eventually. It's like, all right, this is actually like pretty good design on top of a pretty bad system. Yeah. I will say, I mean, it is, again, the natural language capabilities. That is what AI is best at. So it understands me a lot better. And then ChatGPT is familiar. I don't know if it's hallucinating at me, but like it's all happening and there's some guardrails in there. To me, the ones that I...

I don't think those are the Apple intelligence is here features that now Siri can talk to chat CBD. I think that is table stakes and it's kind of weird that it wasn't there from the jump. The ones that I think are big are image playground, Jen Moji and visual intelligence, right? Where the phone gets multimodal and it can now talk in pictures like both in and out.

But I like Genmoji is just purely silly to me. Image playground is like 50 steps behind, although that's probably the safest way for it to be. Yeah. That seems to be deliberate. And visual intelligence is, you know, it's just the same thing we're talking about. We're like, look at, is this a flower? And it's like, it sure is. And like,

The next turn isn't quite there yet. But like, maybe I'm just being a hater. Well, A, you're definitely just being a hater. That's certainly true. I'm in a real salty mood. You are. I can tell. I'm enjoying this very much. You're like, what if everything was awful and woke and I hated it? Look, I know where the views are. I don't know. Like, the Genmoji thing is one that I'm really torn on because I feel like we've been covering Apple's weird ideas about emoji for like the whole life of The Verge.

And I don't think I have ever once earnestly sent or received one of Apple's weird emoji other than like to goof on another tech reporter. I think if like, if I just made a Jen emoji and sent it to my wife, I,

She would like call the police. Like it would be so weird as a thing to just do. Wait, has your kid discovered that you can hold on a face in a photo and make a sticker of a face? No, thank God. Because I have a camera roll full of those or wherever they are in the file system. Like just billions of stickers, like left and right. And those are very funny to send around because they're horrifying. Sure.

But that's very much like we use the walkie talkie feature on Apple watches and we're the only people who use it with a walkie talkie feature on Apple watches. And I just feel like all of this stuff, image playground, gen Moji, like, and this is true of lots of other people's products too, is like, I'm convinced that the, all of the AI image generation stuff is more of a meme than a product at this point. And I just don't, I don't know that there are actual reasons that they exist other than somebody made them. Yeah.

I don't know. Richard, am I crazy? Are you just like sending Genmojis to all your pals every day? I am not. I unfortunately do not have a device ready to run Apple intelligence in my home. I probably never will because I do not want to send Genmoji. I don't want to do this. With Gemini 2.0, you're going to be able to generate some weird stuff too. So get ready. When you open up Instagram, I just want to like search for an account and oh, now I'm in the AI window. Thanks. I really do hate that.

But it's bizarre because I've never wanted to generate an emoji. I don't know if all of the emoji that are in the phone exist and cover every feeling I've ever had. But whatever feelings I have that aren't covered by those emoji,

I'm not in touch with you. I just haven't considered it. It's not something that's just dying to get out of me. I must find the zebra of many colors with anatomical incorrectness that represents how I'm feeling right now. I haven't gotten there yet. I think it's very important for everyone listening to this right now who has access to Apple Intelligence to pull over in their cars,

Send in that prompt and then send us whatever it generates. Like I, I, I issue a lot of instructions on the show. Ask your phone to make a zebra of many colors. That's anatomically incorrect. That is expresses a feeling that you don't know you're having or whatever it is that Richard just said. And just let us know what it gives you.

I'm dying to know. I feel good about that. Look, I think ultimately Apple wants it to seem like the phone is very capable and AI makes, whether or not it actually is capable, AI makes it seem like the phone can do more stuff than it could do yesterday. Right? Now you can press camera control on your iPhone 16 and look at something and it'll be like, here's a flower. You, it can write for you in a way that I think those ads are doing it a massive disservice to because it's

It just seems like everyone's a jerk at work. Yeah, every one of those ads is like, hey, dumbass. Would it seem like not such a dumbass? They have a real problem with those ads because they have to find something that is so important that you have to send the message, but also not so important that you need to write it yourself.

And what is that? Like, how do you, how do you find that you could conceivably be hired for it? Even though you're dumb as a rock, the one where the guy rolls his chair out of the meeting to summarize whatever rolls back in the meeting. I get it. It's a very funny. It is, is, uh, it is attention getting, which is the purpose of advertising, but it's like, Oh, that sucks. Oh,

I would like get out of this meeting. Like, I don't need to waste my time. If you haven't done the reading, um, you can tell what our meetings like, uh, I, I just say like, they've made the phone slightly more capable, but this big Apple intelligence is here a moment.

I don't know. It's not the same as the chat GPT moment that everyone talks about. Like, your mind is not expanded because the messages notifications are 8% better than they were in iOS 18.1. If there's going to be that thing anytime soon, it's going to be agents. And even Apple agrees with that, right? Like, you talk about the intense stuff that they're starting to build into iOS that's going to get Siri more access to apps so it can actually go and, like,

get information and do things inside of those apps. Like that's what everyone is trying to build. And if somebody gets it right in a way that is useful and doesn't cause a culture war, Patel, that's going to be the next thing because those will be the moments where people are like, oh, I now understand a thing I can ask this to do that is useful and new.

The culture where I just want to point out is an AR. It's in visual intelligence. I see. Okay. App intense is money, right? Because Apple is counting on a bunch of app developers to do what they say, which they usually do because their customers are on the iPhone and they're in the app store. And to say, you have to do app intense, just like they said to a bunch of app developers, you have to do in-app purchases so we can take our 30% cut. That's how you get a Siri. They can actually use the apps that you have to build the features in. A bunch of developers are going to say no.

I don't know why DoorDash would say yes to that. The answer has to be money. And like, no one has figured out that money yet. And I, I, I do just keep picking on DoorDash. If anyone from DoorDash wants to come on and tell me that they're going to happily let everyone disintermediate their service, like that's great. But,

I just look at that part and put any company in that mix. Well, look at Netflix. Yeah. You don't have to participate if you have your own audience. And what they've done with the Apple TV and vision pro. If you're big enough, you can just say no. Yeah. So I, I just see that coming, but everyone wants to build these agents and the, the systems they want those agents to use do not have to play ball. And we just haven't sorted that out. All right. So that's 18.2. People get it. Let us know anything. Send us your multicolor emotional zebra.

Please. Let's talk about Sora. Sora is fascinating to me in that it doesn't appear to be great, but it's super good enough to be ultra interesting. Yeah, that's about my read of it. I mean, I think the Sora is so bizarre because it

is another one on the list of things that open AI is like, this is too good. We should barely even be allowed to launch this. We're desperately afraid of all the things that this will visit upon the world if we release it. And then there's just like here, it's a person doing gymnastics whose body explodes in a thousand directions every time they do a flip. And it's like, this is another one where I'm like, does this exist for any reason other than so that people will post about it on the internet? I

I sincerely don't know. Also, it's not exactly available. They very quickly ran out of capacity and stopped accepting signups. Yeah, wait, can you explain to me what happened here? I was away from the internet for several hours and it went from Sora is available to Sora is not available while I was gone. What happened? Yeah, that's pretty much what happened. It was like a shoe release. Like, oh, it's out. Now it's not. And then Sam Altman was like, ah, you know, we just underestimated the capacity and we'll be opening it up.

at some point. So how much capacity do they actually have? How much capacity can they support to run this thing? We don't know. You do have to pay to use it. You have to have either the $20 or $200 plan. If you have the $200 per month ChatGPT Pro plan and you've got access, then you can make 1080p videos that are up to 20 seconds long. So, I mean, pretty good deal. I think like 20 seconds of 1080p video for just $200 a month.

I don't understand why everyone doesn't have two accounts. $200 a month on the ChatGPT Pro plan. I mean, you can use on the ChatGPT. That's the one they just announced. Yeah. Yeah. You can use the ChatGPT Plus plan, but like that's a 720p five seconds. I mean, who's going to do that? That's nothing. You can't destabilize one country with five seconds of 720p video. No.

I mean, look, I watched Marques Brownlee's video about Sora. The most fascinating piece about it is when he asked it to make a tech review video and then it had his plant in it because it has clearly trained on his videos without any permission. What are we doing, guys? Like, I can't tell if Google's going to file that lawsuit or the YouTubers will file that lawsuit, but that lawsuit's coming. Just from that alone, that lawsuit's coming. I mean, yeah, there is so much...

obvious, clear, clean evidence that a lot of open AI is just built on YouTube. It's pretty horrifying. Also, Richard, I would point out you forgot one key difference between the $20 and the $200 chat GPT subscription, which is that if you pay more, you can download the videos without any watermarks. Oh, good. You can definitely destabilize a country if you pay more then. That's...

That's worth $200. Marquez also said something really interesting in his video that I have just been ruminating about ever since he said, I had it make a bunch of CCTV footage because it's so bad that people would assume it's real. Like he intentionally generated footage that was already low quality of like cars driving down the street, but in like black and white security camera footage. And that's like super interesting, right? Because they are 20 second clips with no sound.

And people experience a lot of grainy, low quality clips with no sound that they take to be reality in a lot of contexts, particularly, Hey, we found some security frame of footage or we found some ring footage or whatever it is. So you can pay to get rid of,

The visible watermark you pay to your martial mother, you just crop it, whatever. Are they doing C2PA, the fancy embedded one? Yes. OpenAI says that videos generated with Sora will have both visible watermarks, if you have the cheap version, and C2PA metadata to indicate that they're made with AI. We hope that that works in the

Save civilization. I will say we've written a lot about C2PA and the content authenticity initiative. Jess Weatherbed has done incredible work just trying to understand what it is and how it will play out and whether it will actually mean anything to anyone. And every time we write about it, we get a bunch of angry notes from readers who are, who have been deep in it, who are like, this is a total fraud. Like this will never work the way you want it to for infinity reasons. Um, but it's the thing that we have.

And that's why I keep writing about it because there's nothing else. There's no other choice. There isn't like market competition to this metadata standard. Like there's this one. And Google has chosen to use it. It sounds like OpenAI is using it. Adobe is using it. The platforms have to start using it.

Right. The YouTubes and the Facebooks and the TikToks have to agree to display this metadata and they are not uniformly agreeing to it. And Apple hasn't chosen to use it on the creation side. So we're just all those platforms are furiously building AI tools of their own and have exactly no incentive to point out when things are made with AI. Like I just I don't.

Who benefits from doing that other than regular people and thus, why would they do it? Yeah. We're just headed towards, I think where we're going to end up with is a series of more closed platforms that basically guarantee you stuff from like real people and then open platforms where it's a free for all. And maybe that's going to be good. Like, honestly, maybe it's good to be like on, if I open this app,

I know it's a bunch of like Hollywood movie directors who have actually made the thing. And if I open this app, who knows what random AI generated slop I will get. And like, maybe that's fine. Maybe that's the thing that breaks apart the internet in that way, or like causes that fragmentation. I just, I don't know, but it's, we're just at a point now where, uh,

All of the channels we're used to are going to get completely infused with AI generated content. I think the only problem with that outcome is I think the audience split between those two

uh, different kinds of platforms would not be what you hope it would be. Oh, I think, yeah. One will be expensive. Yeah. One will have a small audience and cost money like that. That, that seems correct. But even now in Tik TOK, I see videos and it just like silly stuff. It's like, here's a big wave crashing over a building and it's like pretty obviously real. And then all the comments are like, this is AI. Oh yeah. And we've just destabilized that entire information economy. Yeah. LeBron was never dunked on. That was AI. Yeah.

A little more. There's more AI stuff. Tries to be launched. Canvas view. What's going on there?

Canvas is basically the slightly more sort of interactive thing that you can do where like if you have it write something for you, it can display the thing that it's making next to the conversation you're having about it. And you can sort of edit the thing in real time with the bot. This is the thing that Claude has been doing, Anthropics Bot, for a while. And I think it's actually very clever, right? Like if you allow for the possibility that maybe...

generating several paragraphs of text is the best thing a chat bot can do. This is actually a pretty good version of that interaction where like you can have it do a thing and it generates it in the right column. And then you can, you can go back and forth with the bot, tuning it and changing it and you can make changes to it. And it becomes this sort of

interactive thing rather than just blocks of messages, which is what we've had before. This is like the big trend here with all these chatbots is everyone has run up against how much you can do inside of a back and forth messaging system. So they're all trying to tack on little bits of UI and little bits of features that don't feel like that.

but still work inside that kind of basic construct. So I think in that case, like canvas is, is a good idea, whether you're, you know, writing stuff or writing code or whatever. It's been, it's been out, it's been tested for a while. I haven't heard a ton about it, but the people I've heard from seem to like it. So it seems good. This thing you're saying about UI across all of these products is super interesting to me. Like,

Chat GPT was the moment, right? Everyone talks about it. Like my eyes were opened. I could see the world in color. And then like two years later, we're like, actually that interface isn't, but we have to build all these other kinds of products to make this useful. And the arms race of like, what will the chat bots themselves look like? Fundamentally, I think they're going to get subsumed into operating systems the way that Apple is subsuming them into Siri. And then there'll be more specialized for other tasks. That's my guess.

But it's interesting to see them rev it. I think that's probably right. I mean, and I think like the way that I've come to frame the way that I'm thinking about AI for next year is the last two years have been like running towards the point of diminishing returns for the actual underlying technology. And we're there, right? This stuff will keep getting better. There might be some incredible physics miracle that changes how all of this stuff works. But on the path that we're on,

This stuff is going to keep getting better, but much slower than it has before. And so now the question is, is any of this useful for anything? And we have not had to ask that question because we've been on basically a two-year run for all these companies of novelty. And they've made it a really long way just by finding slightly new seeming things to do.

And just the fact that I could take a picture and say, is this a flower? And it would say, yes, it's a flower. It was like novel and cool and exciting. And we are at the end of that moment. And I think you're starting to see it, like the disillusionment with what all these things are attempting to do and whether they're actually good is coming and it's here. And so the question now is like,

Let's assume the technology is not going to get several orders of magnitude better next year. What else is there to do with the technology that we have, is the question. Okay, so here's my answer. It's the last one of this section. The most useful thing I've seen with AI yet. Reddit built Reddit Answers.

which lets you just search Reddit without having to go through Google. And it has AI in it. So we'll summarize a bunch of Reddit threads. So when you search for whatever, it'll show you a bunch of Reddit answers, but summarize them with AI. That might be the best thing yet. This might be the best AI feature of all time. I haven't used it. Jay Peters used it a little bit and had, I would say, sort of mixed experiences with it. But the idea of being able to kind of poll all of Reddit with a question that you have

Sounds amazing. And it doesn't work quite as well, even in Google search, which made this very expensive deal with Reddit to get the data because it just, it just isn't compiling them the same way. But like, if I could just quickly be like, okay, what, what baby gear does everybody on Reddit think you should buy? Like that, that information exists inside of Reddit. And if it can go and actually like compile all of that stuff in a way that works and makes sense, then it's a good thing.

kick ass dude i will use that all the time okay now that i've said that i do agree this is very good there of course the supplied examples are what does reddit think happened on january 6th the the the one um uh jay put in was tips for flying with a baby for the first time pretty good like a perfect reddit query right like and and honestly i what i want out of that reddit experience is like give me 5 000 stories about this

Right. And so it summarized everyone's stories. And the answers were consider car seats, fly in first class. Some parents find the extra space in first class helpful. Solid. Cool. I personally find my private shop very helpful with our baby. Feed them during takeoff or landing. Classic advice, by the way. And then bring plenty of snacks and drinks to keep your baby occupied. And it's like, well, that's just stuff. Okay, wait.

This is very funny. I might just immediately take back everything I just said. Because if you go down to the next screenshot, which is the Google searches, the stuff in the previews from the Google searches is dramatically better. Like, let me just read you some of this stuff. It says change diaper before getting on the plane. Don't board first.

Good tips. Being strapped into the car seat is the safest way for baby to fly. Good tip. Baby will need to be in a car seat. Baby will also need to be held by you the entire flight. That seems confusing. You're holding the car seat? Holding the car seat. Take some sanitizing wipes to clean the areas and plane your baby will be touching. Have a light pad you can use for changing a plane. All of that is more useful than this stuff. By the way, can I say none of that is the most useful tip that we found with

with Max, which is just bring five packs of differently colored post-it notes and let her go crazy. Oh, that's good.

Mine was just snacks. We bought, I would say, 40 different snacks. It's not a parenting podcast. I'm just saying it's just funny, right? Because if you take the world's body of information and read it, which is a lot of personal experiences, which are kind of useful and like for this kind of thing where there's not an answer. And then you try to shove it into the form of this is the right answer. You end up with put baby in chair.

But that's the problem that we've run into with Apple intelligence, like notification summaries. Like notifications were already summarized. Like the people who wrote the post already wrote them to be read by someone who wasn't really paying attention and didn't have a lot of time. And they laid it out in a way. They were upvoted because they were written well and they got you the information they needed. You didn't need to summarize them and extract further value out of one particular line. You needed the whole thing. We did it the wrong way.

I mean, this is the ongoing... It's not quite context collapse, but it is...

just the pure like lowest common denominator ism of AI, which is that if you take everything and you try to shove it into two sentences, you're going to get two incredibly uninteresting sentences. And I think what we've seen from some others, like I think notebook LM from Google does a pretty good job of this, of basically it goes and find stuff and then just hands you a bunch of sources, right? Like what I want from this Reddit thing is not

pat answers in paragraphs, I want a bunch of links to like, here are the 10, like you said, Nila, here are the 10 most loved and contentious and funny responses that we've ever gotten to this question that everybody replied to, oh my God, I tried this and it worked. That's how you do this right. Not by saying, here is the rough summary of 12,000 Reddit posts. And that's what too many of these services are trying to do. Yeah. That said, I do think Reddit is...

It's where I see the most opportunity because Reddit is vast, right? And you can burn a lot of time hearing every version of everything on Reddit and having a tool that helps you like get through that better is kind of interesting. I'm just it. Yeah, it's like that's one example. I'm sure there are other ones, but the glory of Reddit is that it's a bunch of people telling you what they experienced. It's very rarely that there's a right answer.

Right. It's just like, I'm going to synthesize all this information from all these people are telling me about whatever, or just talking about their fandom or whatever it is. And that makes me feel like part of a community. It makes me feel, um, it makes me feel like part of a community. It makes me feel validated that like, I might have some experiences too. Not I need it. I just shove her in a car seat. Uh, okay. We got to take a break. We're going to be back with lightning round. Boy, this is, this is an action packed lightning round. We'll be right back.

Support for the Verge cast comes from Stripe. Payment management software isn't something your customers think about often. They see your product, they want to buy it, and then they buy it.

That's about as complex as it gets. But under the hood of that process, there's a lot of really complicated things happening that have to go right in order for that sale to go through. Stripe handles the complexity of financial infrastructure, offering a seamless experience for business owners and their customers. For example, Stripe can make sure that your customers see their currency and preferred payment method when they shop. So checking out never feels like a chore.

Stripe is a payments and billing platform supporting millions of businesses around the world, including companies like Uber, BMW, and DoorDash.

Stripe has helped countless startups and established companies alike reach their growth goals, making progress on their missions, and reach more customers globally. That platform offers a suite of specialized features and tools to power businesses of all sizes, like Stripe Billing, which makes it easy to handle subscription-based charges, invoicing, and all recurring revenue management needs. Learn how Stripe helps companies of all sizes make progress at stripe.com. That's stripe.com to learn more. Stripe.

Make progress.

Support for The Verge Cast comes from The Home Depot. It's almost the holidays, and whether you're planning to travel or host, it's always a good idea to have an extra layer of safety and security to help ease your mind. And now, with help from The Home Depot, you can stay connected and protected with the convenience of smart home security products. The Home Depot offers a wide selection of products that afford you easy control and automation of your home. With top smart home brands like Ring, Google, Wyze, and more.

From smart home cameras with 4K surveillance to doorbells that can be operated with your smartphone, the Home Depot has it all. Whether you're an expert or you're new to smart home tech, the Home Depot can help you find what works for you and your home. Visit the Home Depot online or in-store and purchase your smart home products to give you some peace of mind this holiday season. Because smart homes start at the Home Depot. Support for The Verge Cast comes from Polestar.

Polestar is looking to usher in a new generation of electric SUVs, one that relies on technology so integrated you forget it's even there. And they're starting with Polestar 3, their luxury electric SUV. Polestar 3 features an aerodynamic exterior and a Scandinavian minimalist interior.

It has the ability to go from 0 to 60 in as little as 4.8 seconds and get an EPA-estimated range of up to 315 miles per charge. Polestar 3 even lets you optimize the powertrain between performance or range mode, depending on your drive's needs. You can also say goodbye to that cluttered dashboard because this car shows you everything you want to know and nothing you don't.

You can even have Google turn on your favorite podcast, we're hoping it's this one, and be immersed in 3D surround sound by Bowers & Wilkins. Polestar 3 has a lot to offer, so if you're a driver who doesn't want to choose between spacious comfort and the agile handling of a sports car, then this might be the car for you. You can test it out for yourself at your local Polestar space. Book a test drive for Polestar 3 at Polestar.com.

Thank you.

Glam Bag is a curated selection of the best of makeup, skincare, and hair care products, all curated by Ipsy's experts. Every month's delivery brings fresh selections, including top brands like Benefit, e.l.f., Bare Minerals, Dry Bar, Makeup Forever, and many more. And the products in each Glam Bag are worth much more than the subscription cost. So it's a high-end beauty experience without the high price tag.

This holiday season, you don't need to spend time researching to find one gift to satisfy your special person. Instead, you can continue to surprise and delight them with Ipsy. They can redeem the subscription whenever they choose, and who knows, they might find their next holy grail product in a glam bag. Ipsy is the gift that keeps delighting month after month. Give the gift of beauty for three months starting at just $50 at ipsy.com slash gift.

♪♪♪

All right. We're back with the lightning round. Liam, who is the lightning round sponsor? This week's lightning round is presented by Amazon Q, the new generative AI assistant from AWS. Oh, it's so good. Have we started adding the cash register sound effect? I think we only did it on the first one since you asked for the effect, but if you, if you want that to be a regular thing, we could...

I feel like we got in a lot of trouble with an episode recently by asking for sound effects that we didn't deliver. Also, if you'd like to make a Genmoji of Nilay being showered in cash and send it to me, I'm good with that. Right next to our emotional zebra. All right, lightning round. David, you wrote about this this week. YouTube is still growing fast on TVs and living room. What's going on here? Yeah, so I would say statistically, if you are...

absorbing this podcast right now. I don't even know what word to use anymore. If you are experiencing this podcast...

uh there is an increasingly large chance you're watching us on your television in your living room uh if you are hello oh that's only if you're watching this on some janky side lit lcd i still love you you should actually watch it watch it like several times actually yeah watch it on all of your different tvs just to check the resolution differences i would say um but no basically youtube put out a bunch of stats this week about how fast its living room stuff is growing uh

400 million hours a month of podcast viewing is happening on TVs. Sports stuff is way up. They launched this new feature that I think is kind of cool. It's called watch with, uh,

Uh, and essentially what they discovered is that a lot of people are watching a sporting event with live commentary from somebody streaming and talking about it. And they're now smushing those things together. So creators are going to be able to actually like do their own commentary over top of something that they're watching. And they're starting with sports, but also made it pretty clear that that's going to come to other things. So like Kurt Wilms, the guy I was talking to who runs product for YouTube's living room stuff.

specifically mentioned, like, what about an Apple keynote where there are all these creators who want to talk about it and comment on it? That's just a thing we can offer, that you can get all these different commentary streams. I think that's very cool. I'm sure Apple will have some very interesting thoughts about that. Yeah, what could possibly go wrong? Who's going to be mad about new commentary on every movie that exists in the world? But, uh...

More broadly, YouTube is now huge on TVs and has undergone this shift, I think, really in the last year. I follow this company pretty closely. And they used to talk about the TVs as just kind of another place people watched YouTube. And now they're really thinking about it as a primary platform for YouTube. And they're building features for the TV industry.

kind of for the first time. Like they have this shows thing where you can put up a bunch of videos in seasons and episodes that makes it feel more like a streaming service. They're doing the watch with stuff. They're doing a lot of work to sync your phone with your TV. Like YouTube has always kind of wanted to be Netflix in a certain way. And I think we're starting to see it push to be a little more premium, a little higher end, a little more TV centric.

They're convinced they can do this without killing YouTube as a product elsewhere. But that tension feels very real to me. And I think it's really fascinating. Yeah. Can I say one thing on YouTube very quickly? We've talked enough about our business in the past couple weeks. I don't want to overdo it. But I have a story on the website. This week, a lot of people complain that our YouTube videos and our website don't let you click the title to go to YouTube. And everyone thinks that's our fault.

Because it makes sense that the publisher would want you to keep you on a website, but

It just, it makes sense to people. And because forever, every YouTube video on the internet, you've been able to, you, you hover over it, you click the title and it takes you to, and in particular on phone, you want to open the YouTube app. There's all this stuff you want to do. I, I hope people know that the verge is like very pro links. We have links everywhere, like links on the homepage all the time. Um, that's not our fault. It is a YouTube decision to disable those links. And if we want to re-enable them, we have to make less money on our YouTube videos. You can read the whole story. Uh,

I was just very annoyed that I spent months being like, can you turn on these links back on? And the answer at the end of the day was a hard no, unless we choose to either not use YouTube at all or make less money. And I'm not really part of the business side. Maybe the business side make a different decision. But I was like, that sucks. I'm just going to tell everyone about this. And all the YouTube people just made a very sad face at me. If the internet goes back to every publisher having their own awful bespoke video player again, I'm going to be so angry.

Like feel however you want about YouTube. It's a good video player and the internet was filled with bad video players for so long. By the way, the links worked and we made more money since 2016. It was all fine. And then in the start of this year, they changed it to quote remove their branding and they think the link to YouTube counts as branding. That's dumb.

It's dumb. Let me tell you how much of a brat I've been for the past two and a half months. And the answer is a hard no. And I was like, I'm going to write me. I'm still around. Can Vimeo fix this? Sure. All right. Here's a really weird one. I'm obsessed with this one. Instagram is going to let creators test reels on people that don't follow them. So like,

Adam and Sarah made this really should go watch it. It's very interesting as a, as a sociological document of how Adam thinks Instagram creators think about Instagram, like watch this video through that lens.

He's like, I know a lot of you are stressed when you upload a reel that it won't perform. So now we're going to let you upload a reel. We won't show it to anyone who follows you. We'll show it to people who don't follow you so you can see how it'll do before you publish it to everyone. And it's like, that's not how, that's not how you should feel about your creative work. Right? Like I'm going to focus group. My video before I show it to the people who follow me is a weird thing.

It is correct. I think if you're making like advertising, you run the pressure washing business, like, is my pressure washing business going to do better? Cause I made, maybe that makes sense, but that's what I'm saying. Adam thinks everyone is making commercial videos.

And they're trying to optimize them for reach as opposed to making stuff on Instagram. Richard, what do you make of this? This makes perfect sense to me. I think it explains, like you said, it explains a lot about the way that Adam Mosseri and Instagram kind of see the world, especially what we've seen with threads, for example. The way that you could have a post that you would put up and it would go viral with strangers, but

but not be seen by the people who follow you. And in particular on threads, go viral with strangers who don't like what you said. It was very, very, very good at identifying people who do not like what you said and making sure that they see your posts.

But this is kind of just the way that they see it. Okay, so you made something you would love to be seen by whoever, random people, not your community, because you would be deeming them in your channel if you wanted that. - No, but they're saying you don't want to blow it with the people who follow you. So we'll like focus group it with people who don't. And then you can decide if that data makes you confident to show that people that follow you.

Maybe I think that's interesting. I'm like, that is legitimately one of the most novel features a social network has released in forever. Super interesting. I think the idea that you kind of can't trust the people who follow you to judge whether or not your content is good. It's just a really weird way to think about things. And that might permanently downrank you in the algorithm. Like that's what I mean. Like the, when I say watch this video from the perspective of how does Adam think about creators thinking about Instagram? It's just purely commercial.

It's like, what you want as a creator is to do great in our algorithm. So here's some tools that will help you do great in the algorithm. And it's like, actually what I want as a creator is to make art.

Like those are different things, like wildly different things. This is going to sound more cynical than I mean it, but I would, I would take pretty strong bets on that. More people are interested in making money than making art. Oh yeah. But that's what I think those platforms where we are. I think creator platforms are commercialized in that specific way. People want to win them. And here's a feature. And you know, he even says that video, we made this with creators. Like what we've gotten a lot of feedback that we've been working with creators, um,

But there's just something very commercial about everything that's happening there that is very far away from... This was a photo-sharing website for me to talk to my friends. Yeah. Go watch the video. There's something really specifically Instagram-y about this that I find really fascinating. Richard, like you were saying, what this says to me is that you are...

always on a knife's edge of losing everything on Instagram. In the way that like on TikTok, everything you make is like a new pull at the slot machine, right? Like Instagram cares so much more about who follows you and who you follow than TikTok does. TikTok is just happy to shove stuff into the algorithm and see what works and you'll randomly go viral and sometimes you won't. Instagram lets you build something

that is like much more sort of understandable by you as the creator. But also if you make two videos in a row that your audience doesn't like and respond, doesn't respond to your toast. And that this is like every creator's desperate fear is if you miss once it can, it can kill the whole thing for you. And, and what this is, is Adam Asari basically saying, yeah, that's true. So we're going to give you more tools to make sure you never do that. It's weird. Yeah. Weird. Uh, speaking of, it can all vanish an instant, uh, Tik TOK lost its court case, uh,

challenging the law that would force it to either ban itself or be sold. Lauren Feiner wrote up the decision on that case. Complicated, but the court basically said the Congress made a national security decision. We're not going to override it. Many complicated First Amendment questions. We'll deal with them probably in an indicator episode down the line. But,

Donald Trump, incoming president of the United States, asked about this by Kristen Welker on NBC News in an hour and a half long interview. I'm just going to read this quote because it's a lot of words from Trump. I use TikTok very successfully in my campaign. I have a man named TikTok Jack. He was very effective, obviously, because I won the youth vote by 30%, which he did not do.

do, but whatever. I just wanted to say TikTok Jack. Anyway, TikTok Jack, email us. I need to know everything about you. And then he said, I use TikTok. So I can't really, you know, I can't totally hate it. It was very effective, but I will say this. If you do that, meaning ban it, something else will come along and take its place. And maybe that's unfair. And what the judge actually said was you can't have Chinese companies. They have the right to ban it. If you can prove that Chinese companies own it, that's what the judge actually said.

She has to follow up. She said, will you protect TikTok? And he said, I'm going to try and make it so that other companies don't become an even bigger monopoly. And specifically what he means is Facebook. He has often complained that Facebook would just become a bigger monopoly if TikTok is banned. And then during the campaign, he obviously said to a bunch of young people, I will protect TikTok. Biden wants to ban TikTok, I'll protect it. This is, in my mind, a massive walkback, right? All he's saying is, I don't want Facebook to be a monopoly. Not, I will protect ByteDance owning TikTok in America.

And I'm, I'm just betting. I think we we've done this a million times now, David, like who's going to buy TikTok because that's the out. The out is he negotiates a deal for Amazon or Walmart or subway to buy TikTok. Uh, and he gets to say he saved it. I made a great deal to save TikTok. Aren't you proud of me? I'm doing what I said I would do. And I'm, I just, you can watch the clip. We have it on the website. It feels very much like that is the door he's opening.

I mean, it's the escape hatch, right? It's the closest thing he's going to find to a win-win at the end of this. He gets an America first win. He gets TikTok to still be here. He gets to, like, poke a knife at Mark Zuckerberg. Time's running out, though. Ban goes into effect on January 19th. TikTok has filed with the Supreme Court to appeal. We'll see what happens. Notably, January 19th is the day before January 20th, which is when he takes office. So...

Wait, Neil and I have been on the record about what we think is going to happen many times. Richard, you have to make a prediction right now what happens on January 19th. Oh, TikTok actually turns into a new app called TalkTik. It won't be owned in China, it'll be owned in like Malaysia or something. But, you know, by a very mysterious company that just happened to come into a lot of money. Who knows where it came from.

Uh, next one, Chris Welch reviewed the Sonos Arc Ultra. He says the hardware is great, particularly the new transducers from Mate. Uh, it's a company that's on a spot to make smaller transducers and make more bass. Loves those. App is still a little, little something. Richard, what do you think? Uh, I don't know. I mean, it's, it's Sonos and they still have the problem of the app that they messed up and made everyone mad. So, uh, how's that going? Medium. I would say the answer is medium.

Um, but I will say that as a soundbar seems very cool. If I, Chris was basically like, this is worth upgrading, which upgrading from one expensive soundbar to another expensive soundbar is almost not like no one has that thought. Like I'm a person who likes to buy speakers and I'm never like, I should upgrade a soundbar. Like, uh, and Chris thought it was worth upgrading, which is fascinating. So hopefully, hopefully this is beginning to turn around. Uh, this was big news this week.

Uh, GM shut down the cruise robo taxi service, laid off a bunch of people. The former CEO of cruise, uh, just flat out posted GM as a stupid company, which is great, but it feels like used was dummies, which I like very much. Dummies is a, is a surprisingly powerful, powerful, mean thing to say to somebody. I don't know what David, there's just like a lot of action in this world. And obviously GM is like up and down. What do you, what do you think is going on here?

I think it just is starting to seem like all of these car companies got out over their skis on technological revolutions, right? There was the sense that EVs and self-driving were going to happen

really fast and that they were going to be immediately mainstream and they were going to become a big business. And everybody got really excited about the idea of robo taxis because then you have a really interesting, like diversified business from your own vehicles. Like you can see how they get there.

But we're not there. And I and like, unless you're Tesla, and people just keep giving you stock price money with which to have weird ideas about where to go from here, you're still fundamentally running a car business, which is not getting higher margin, it's not getting less complex.

You still have to make the things that people are buying now. And it just feels like one by one, these car companies are starting to say, okay, we made these big giant future bets that just aren't coming true as quickly as we thought. And we just have to get out of it. I also think a lot of people who cover this space more closely than I do think there is something weird going on with this one in particular, that the way GM is handling this suggests that something else is going on inside of cruise that, um,

it basically just like pulled the plug all at once, which is a very odd thing to do when you're this kind of company that is this invested in cruise. But the macro thing actually kind of makes sense to me. It's just everybody made a bet 10 years ago now that they thought was going to be true in five years. And now it looks like it might be more like 25. I think it's just hard to lose money the way Google has chosen to lose money on Waymo for this whole time. Yeah. Right. I mean, that's just Google's like, we're just going to lose money until this thing can make it through Phoenix. Right.

All right, we're going to keep losing money until we can make it through Austin. And it's just like... Yeah, GM just doesn't, literally doesn't have that money. Yeah, it's just not a thing. Richard, you're the one who put quantum computing on this list. Google reveals quantum computing chip with breakthrough achievements. Tell me why this one made Richard happy. Because most of the other stuff did not. I fully understand quantum computing. I understand everything about it, actually. The quantum computing chip... Richard is currently both here and not here.

As I have said many times, my coin, RJCC coin is quantum locked. It is both launched and mint and pre-mint at all times. That's how it works. That's how quantum works. That's what we do here at RJCC coin. And we're going to do it on Willow because it can perform a task in five minutes that would take a supercomputer 10 septillion years to complete, which may be evidence that we live in a simulation. Um,

or a multiverse, maybe or maybe not. I don't really know what any of these words mean, but it happened. They have a chip, they're doing a thing. And now they're trying to find, basically, now that they've done this thing, they're trying to find something to do with it that you will actually find a use for so they could prove how fast it is. Because right now, all the stuff they can do is theoretical. The answer is breaking crypto.

RJCC coin. That's what's going to happen. You're already in. It's quantum. You're already in it. You already have it. I have three very brief things to say about this whole thing. One, everybody should read the research paper that Google did because even Google's own researchers are basically like, this thing we did is cool, but I don't know what it is that we just did here. That's all quantum computing. Yeah, they're like, we did an amazing...

unprecedented mathematical calculation that no one cares about. It has no bearing on the real world. Like they, they say that it's amazing that we did it. We don't know why we did it or what it actually accomplishes. And then they're like, the main thing that we've solved here and our real breakthrough is that we've made it make fewer errors. Yeah. Like, like imagine if your computer was just like, sometimes when you try to move a file, like,

It just explodes. But now our computer does that less. It still does it. It's very good. Less. I'm just saying the main purpose of quantum computing is going to be to break cryptography in specific ways.

You're ready. Yeah. You're ready. But then the third thing is that they just very casually are like, this might be evidence that we're living in a simulation. And that's just like the end of it. They just, they don't address that anymore. They just say it. All right. Here's the last one, which is pure evidence. We live in a simulation. You can now buy a Hyundai on Amazon. It's very weird. You can like, it's very weird to go on Amazon and see something listed for a price of $67,000.

But there's no one click buy. It's start. The button is like, start now. And then you get kicked to a dealer. And I feel like we're, we have to buy a car to test out if the dealer actually plays ball. Wait, that's less exciting. So I was really hoping there was going to be like a Amazon warehouse somewhere. That's just full of cars. That's Carvana, right? There's the vending machine. Yeah. The car vending machine. This is there. I think they're just doing some pricing games with Hyundai. Right.

Right, where Hyundai sets a price, Amazon sets a price, and the dealer just delivers the car and everyone gets a cut and is happy. But there are still car dealers involved. And I assure you, at some point, they're going to be like, and we've marked this one up. We filled the tires with nitrogen. So now that's an extra $6,000. Like that's car dealer stuff, right? If anyone wants to buy a car on Amazon and tell us how it goes, I'm dying to know. If you just want to cut me a check for $65,000 so we can buy a car on Amazon.

We will accept that money as well. But I'm kind of dying to do this because I want to see if the dealers play ball. This is so bizarre. It's weird. Amazon will give you a $2,300 gift card if you buy a car through Amazon. It's the future, man. It's like kind of like a tax break, but it isn't. The weird thing is the car only works on Kindles.

But I want them to link up and bring a trailer so I can just go on Amazon and somebody's Mercur XR4 Ti is up there and I'm just like, yeah, buy it now. And then it shows up at the house the next day and I have to explain that.

That's what we need. They bring you the whole thing and then the delivery driver just hands you the keys and walks away. My cousin in New Jersey has absolutely impulse bought a car and bring a trailer and had like a 1980s Mercedes roll up in his driveway. His wife is like, what's going on here? But you can absolutely impulse buy a car and bring a trailer. That's a real thing that happens. All right, we got to get out of here. This is a great one, Richard. Thanks for being here.

And I want to call out one story, which is maybe my favorite story of the month so far. Kristen Radke, our creative director, and Amelia Halliday-Krails, our photographer, went to Pantone's launch of the color of the year, which is called Mocha Moose, which is brown. So they went to a party for the color brown. Kristen was horrified because she was accidentally wore brown on that day. Yeah.

But then she got to the party. Everyone else was wearing brown. It's like a perfect Verge story. Everything's great about it. Go read that story. The photos are amazing. That's it. That's VergeCast. Rock and roll. And that's it for the VergeCast this week. Hey, we'd love to hear from you. Give us a call at 866-VERGE11. The VergeCast is a production of The Verge and Vox Media Podcast Network. Our show is produced by Liam James, Will Poore, and Eric Gomez. And that's it. We'll see you next week.

That's aws.amazon.com slash Q.

Support for this episode comes from AWS. AWS Generative AI gives you the tools to power your business forward with the security and speed of the world's most experienced cloud.