Pod Insights Hard Fork Elon's X Machina + Crypto Orbs + A Visit to Google’s Robot Lab About this episode On Sunday night, a crane arrived in downtown San Francisco to take down the Twitter sign from the company’s office building. The crane’s arrival marked the death of Twitter, the brand, and the start of X, Elon Musk’s everything app. Today, why Elon’s acquisition feels more and more like cultural vandalism and what, if anything, will replace the global town square.
Then, is Sam Altman’s universal basic income cryptocurrency app Worldcoin an iris scanning tool to save humanity, or just another attempt to get rich on crypto?
Plus: a trip to Google’s robotics lab, where artificial intelligence models are creating breakthroughs.
Transcript 11,758 words · 10 speakers
Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner, Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
0:35 Kevin Roose? Do you ever like, when you were edited, when you had an editor, when you worked at a real publication?
0:40 Casey Newton? Yeah.
0:40 Kevin Roose? Did you ever put stuff in a story that you knew was bad just so your editor would cut that and not the thing that you actually wanted to stay in?
0:48 Casey Newton? No, and thank God I never worked somewhere where I had to think like that. Because like, trying to play the nine dimensional chess of like, how do I get something past my editor? Would have driven me out of this business.
0:55 Kevin Roose? I call this my green M and M strategy, where I like, I put something that's like horrible in the story and I just like, leave it there. And then there's something like three paragraphs below that, that I know they're gonna wanna cut, but that I really want them to believe in.
1:08 Casey Newton? Right. Kevin would just be like, admittedly, some have asked whether 9/11 actually happened. And the editor says, you know what? I know how we can save one paragraph worth of space.
1:24 Kevin Roose? I'm Kevin Rus. I'm a tech columnist at The New York Times.
1:26 Casey Newton? I'm Casey Newton from Platformer.
1:27 Kevin Roose? And you're listening to Hartford.
1:29 Casey Newton? This week on the show, X marks the spot where Twitter died. Plus, why iris scanning orbs may be coming to a city near you. And we visit Google's robotics lab, where they are putting AI inside the machine's brain. We'll tell you how many have escaped the lab.
2:00 Kevin Roose? All right, Casey, I have been dying to talk to you about what the hell is going on at the company formerly known as Twitter.
2:09 Casey Newton? Well, Twitter doesn't exist anymore, Kevin. We have to stop using that word. It's no longer an active concern.
2:15 Kevin Roose? Right, so let's just do some chronology up top because I think it's useful for people to understand how quickly and kind of erratically all of this happened. Right, so Saturday night Elon Musk tweets at about 9 o'clock Pacific time, "Soon we shall bid adieu to the Twitter brand and gradually all the birds." Now, we have known for some time that Elon Musk has wanted to rebrand Twitter. He started a couple of holding companies for Twitter called X Holdings. So we kind of knew that something was afoot. But talk to me about how sort of chaotically and quickly this all happened.
2:53 Casey Newton? Well, as you noted, it was all done in a very haphazard way. On Sunday night, they projected a giant X onto Twitter's headquarters on Market Street. And then Monday, a crane showed up and started to take down the Twitter sign.
3:10 Kevin Roose? Right, and actually got stopped by San Francisco police before they could finish. So at least for a little while, the sign on the Twitter building just had an ER left, just was "err."
3:21 Casey Newton? "err," yeah, just "err," which really does capture, I think, a lot of people's feelings about the whole thing. It is, "What did Elon Musk do to that?" "err," hmm, yeah. So anyway, by that point, of course, the product had begun to transform. And so if you opened up the app, all of a sudden, where you used to see the bird, you now saw an X.
3:40 Kevin Roose? Yeah, and at most companies, like a rebranding is something that takes many months. It involves lots of consultants and designers and focus groups.
3:48 Casey Newton? Paying people to do things for you.
3:49 Kevin Roose? Exactly. This was like, you know, let's just throw it up and see what happens. And not only was the product sort of renamed, but actually people started going around inside Twitter's headquarters and like pulling the bird logos off the walls and... And renaming the conference rooms. So my colleagues at the New York Times reported on Monday that the company's conference rooms have been changed to words and phrases with the letter X in them, like exposure, exalt, and sexy with a three instead of an E.
4:22 Casey Newton? I'm so excited to get laid off in the sexy room. That's gotta take the sting out of it.
4:27 Kevin Roose? Basically, any room you're in is the sexy room. But I just like, I think it's really strange. And actually, I had a moment this week where, you know, I was opening up my computer to start my workday like normal.
4:41 Casey Newton? So it's roughly noon, are we talking?
4:43 Kevin Roose? Come on now. So I wake up, I pull up my Chrome, I instinctively go to Twitter. And instead of the like blue bird in my like little browser tab, there's this like black X. And that was the moment for me where it clicked. Like, oh, this is not Twitter anymore.
5:03 Casey Newton? Yeah.
5:03 Kevin Roose? Like the thing, like, I don't know why that was the moment for me. But like, until now, if you didn't really like, think about it too hard, you could kind of just squint and pretend that Twitter was still the old Twitter.
5:15 Casey Newton? And lots of people have been doing that, basically, just trying to pretend all the Elon stuff is noise and I'm just gonna keep tweeting like I've always been tweeting.
5:22 Kevin Roose? Right, and you could kind of fool yourself because some of your friends were still there. The site and the app looked roughly the same. But like, for me, that little black X in the browser bar was the moment where I was like, oh, it's gone. Like, I'm not using Twitter anymore. This is a site that's owned by Elon Musk, who probably does not want me on it. And it just was a moment where it went from feeling like a thing that I have been doing for like more than a decade at this point, when I opened my browser, is no longer a thing that feels good to me.
5:50 Casey Newton? Absolutely. And look, I think Twitter has died a lot of deaths since October, when Elon Musk walked in with that sink. And in fact, I think that day was probably the first of those deaths. I think it died again when he got rid of three quarters of the staff. I think it died as he gradually replaced the old content moderation policies with sort of more than anything goes ethos. But the moment when the website actually just redirects to another name and the app itself has another logo, yeah, Twitter is no more.
6:25 Kevin Roose? So I want to talk about why this is all happening, because it seems like Totally incomprehensible to me. And let's talk first about the official explanation, which is the one that Twitter CEO Linda Iaccarino gave.
6:40 Casey Newton? And it's so beautiful. This is some of the best writing that you'll ever hear. And Kevin, I hope you do it justice when you read it.
6:45 Kevin Roose? OK, let's do a dramatic reading of her tweet.
6:46 Casey Newton? Yeah, OK.
6:47 Kevin Roose? X is the future state of unlimited interactivity centered in audio, video messaging, payments slash banking, creating a global marketplace for ideas, goods, services, and opportunities.
7:02 Casey Newton? I mean, what doesn't X do, you know?
7:04 Kevin Roose? Right. So what is the official explanation that the company is giving for why they are changing the name of the service?
7:10 Casey Newton? Well, going back to before Elon completed the transition, he talked about wanting to make Twitter into a super app, right? So in China, you have WeChat, this app that is your entire digital life. You shop with it, you post on its sort of internal social network, it's your pass on the train, right? And that's a huge opportunity, right? It's essentially, it's like, well, if one app on your phone constituted all of the economic activity that you did, Elon Musk would make a lot of money. So that's what he wanted to do. It will not be possible for him in the United States for a bunch of reasons, but that is the goal, is that X is essentially going to be your one-stop shopping app for everything.
7:51 Kevin Roose? Right. So that's the official explanation that we're getting, is that Twitter is becoming this like, you know, huge super app that's going to do everything from handling your banking information to like, you know, getting you onto the subway, to, you know, paying you for your banger tweets or zits or whatever we're calling them now.
8:10 Casey Newton? Yeah. And by the way, have you uploaded all your banking information to X.com yet? I've been waiting.
8:15 Kevin Roose? I saw the activity over the weekend and just the thoughtfulness of the new X rollout. And I thought, I would like to give that man my social security number and routing number.
8:24 Casey Newton? I called up Wells Fargo and I said, it's over between the two of us. I'm taking my business to a man that gets me.
8:31 Kevin Roose? Okay, so you have a different explanation for why all of this is happening. What is your take on why Elon Musk is doing this?
8:39 Casey Newton? Yeah, so, you know, the way that I have described what he's doing is cultural vandalism, right? I see this as very much a political project and specifically a reactionary political project. And we've talked about that a bit on the show in the past, the way that Elon Musk was really aggrieved by what he saw as a growing sense of entitlement among tech workers in general, but specifically at Twitter. Twitter had endorsed a lot of liberal values and causes like Black Lives Matter. And, you know, I'll just never forget when, after he took over, he opened a closet and he found a bunch of T-shirts that the old Twitter had made that said, "Stay woke on them," which sort of recognized the company's role as a part of the Black Lives Matter movement, right? And to Elon, I mean, he just could
9:24 Casey Newton? not stop making fun of this. Like, this was just so funny to him that, you know, essentially anyone would support the Black Lives Matter movement. But it was the way that he was preening for the camera when he talked about this that sort of let me know, this is the project. This is the point. It is to go into this important cultural institution and to tear it down from the inside and to say, "These values no longer apply. I am going to wipe this thing off the face of the earth." And we eventually got there, and it didn't even take that long.
9:50 Kevin Roose? So you think this is basically like a kamikaze mission, right? He is trying to run Twitter into the ground. My theory about what is happening here is partly what you said, is partly that there's this sort of cultural vandalism happening. But I also wonder if, like, he just is starting to feel like Twitter is too boring and too small to command his attention. Like, he is clearly obsessed with Twitter, the service, but I think as time has gone on, you know, it's been almost a year since he acquired it now. And I think that he is just kind of out of ideas. He tried subscriptions. He tried sharing revenue with creators. He tried longer videos and bringing back the people who'd been banned, like, all these ideas. And nothing has really kind of moved the needle in the direction
10:35 Kevin Roose? that he wants it to go. And so this, to me, feels like kind of an act of surrender. Like, he's just kind of saying, "Well, I can't fix Twitter, so I'm just going to rename it and fold it into something bigger, and maybe you'll forget about how disastrous the past year has been."
10:48 Casey Newton? Well, look, in the run up to him acquiring it, a lot of folks said, you know, he's probably going to get bored of this, right? Because it's actually hell to run a social network. You do spend most of your time having to weigh in on these very difficult questions about content moderation, the business model's a challenge. And I think a lot of folks thought, this guy's got a lot of other things on his plate. He's just not going to want to stick with it for the long haul. He, of course, famously tried to get out of the acquisition, was eventually compelled to buy the thing by a court. And so then he's in charge of it. And as you point out, he tries some things and they don't work. I do think that it is reasonable to assume that in the aftermath of the failure of his original set of ideas for Twitter, he's now beginning to think like, well, what else could I do? I guess I could go dust off this idea that I had in the 90s for a payments app and just
11:33 Casey Newton? hope that the people who are still using Twitter will come along for that ride.
11:37 Kevin Roose? Right.
11:38 Casey Newton? But let's just say, that's a terrible idea.
11:40 Kevin Roose? It's a very bad idea.
11:42 Casey Newton? It's like, you know, it's like if you've acquired the user base of Coke and now you're saying like, and now we're going to sell them mortgages. It's like, well, you can try to do it, but you went about it in the worst way imaginable.
11:54 Kevin Roose? Right. It does not inspire confidence that one would need to like hand over your finances to this man.
12:00 Casey Newton? Right.
12:00 Kevin Roose? So let's talk about these Twitter alternatives.
12:03 Casey Newton? Yeah.
12:03 Kevin Roose? Because, you know, a few weeks ago, we had our special episode with Adam Mosseri, the head of Instagram, right, as they launched Threads.
12:11 Casey Newton? And just to sort of remind everybody what happened, we did a podcast when Threads launched, and within the first week, it had 100 million signups. So we're calling that the hard fork bum. And it could have worked for your product launch, too. Get in touch.
12:25 Kevin Roose? Yeah, we don't promise 100 million users for every product, but maybe 50 million. You know, the forker army is strong out there. So Threads has now been in existence for three weeks. How's it going? Like, what's your impression of whether this is actually the social media app that can replace Twitter?
12:44 Casey Newton? So there's two ways of looking at it. One is you can look at how engagement has been on the app since it launched. And that is a chart that has basically been going down over time. A lot of people opened up the app, they browsed a bit. They saw a lot of posts from brands. They didn't see their friends. They thought this kind of feels like a shopping mall and they bounced. And so some folks who were frankly predisposed to saying this thing is never going to work. It's just a copycat app from a company that I hate. They've been doing some victory laps lately, saying this thing does not have the juice and it's going to flop.
13:17 Kevin Roose? Right. And there is some data to support this. SimilarWeb, which is a company that measures for the popularity of various websites, estimated that Threads' usage has dropped by about 60% compared to its initial kind of surge of interest. So, you know, a bunch of people rushed to sign up. Maybe they saw it on their Instagram. Maybe they heard about it on Hard Fork or somewhere else. They sign up, they go, they maybe post a few things, read a few things, and then they drop off. And that seems to be happening for a lot of people. That's not just like a, like an I hate Meta thing. This is like, it does actually appear to be losing popularity.
13:48 Casey Newton? Yeah. So, I mean, I think that that is true. I also think that that is true of most social networks when they launch. There's this initial rush of attention and then most users churn away. And then it's up to the makers of that social network to figure out, okay, how do we build this thing so that people keep coming back time after time? And that is maybe the thing with which Meta has had the most practice over the years as it has built Facebook, as it has built Instagram, as it has built WhatsApp, right? So I'm not particularly bothered by all that. The more optimistic framing of what's happening at Threads is they launched a bare bones copycat of Twitter. There was so much demand for something like Twitter in the world. They got 100 million users within a few days. And now they have to do the hard work of building out the rest of the features.
14:32 Casey Newton? And so the game is how quickly can they ship the features that will not just make it what Twitter used to be, but actually better than what Twitter ever was? And I think they're well on their way, right? So this week, they shipped something that people like us have been asking for since they launched, which is you can now see a feed of posts just from the users that you follow, right? For the most part, when you open up Twitter, for most of its existence, you would just see stuff from people you follow. That's what made it very useful. Threads opened up to this algorithmic timeline, much more Tic Tac like. It would just sort of show you things that thought you might be interested in.
15:05 Kevin Roose? Which are mostly like Wendy's from my experience or like brands?
15:08 Casey Newton? Who doesn't love a cold Frosty on a summer's day, Kevin? But, you know. So, you know, it has, it still has a lot of levers that it can pull to grow the thing. And I would say that within a couple of weeks, it has just become far and away the most likely thing to replace Twitter in the popular imagination. There are alternatives, but this is the one that I just think has the most momentum.
15:30 Kevin Roose? Yeah, I'm gonna take the other side of this. And partly, that's because I'm one of these people who like, opened threads and used it a bunch for the first couple of days. And then ever since, I just like, it just hasn't been that interesting or compelling to me as a product. And so I've basically churned off at this point. But it's also just like, it is worth reminding people of why Meta has the reputation that it had, why people are skeptical of this company. There was a period of many years where Meta slash Facebook was just pursuing growth at all costs. You know, a lot of that resulted in real concrete harms. There was a genocide in Myanmar. There was ethnic violence in
16:09 Kevin Roose? Sri Lanka. That was a direct result of Facebook's decision to grow into those regions without having the infrastructure that they need to support. And now we know all of this, in part because people kept leaving Facebook and saying, we knew about this at the time. We knew that these harms were present, and we didn't do anything about them because growth was more important. And so I think when I see something like threads, and I see all the Facebook executives, you know, celebrating the growth of this new app that they've built, what I'm thinking is like, what are we going to learn 12, or 18, or 24 months from now about what's going on actually under the hood at threads? And so that's what makes it feel a little bit less fun for me, because I actually still remember
16:50 Kevin Roose? the period from like, 2014 to 2020, where Facebook just was causing real harm all over the world because of this appetite for growth.
16:58 Casey Newton? I think these are great criticisms. I co-sign basically everything that you just said. Meta has a lot to answer for. I've spent a lot of my time in tech journalism, writing about those specific things, right? And I absolutely believe that you and I and everybody else on the Tech Press Corps is going to continue to scrutinize this company as this thing grows and as it... Creates potentially more harms. It's also true that Meta has dismantled some of the integrity teams that it had built over the past several years on the wake of the 2016 election. So I think all of these concerns are valid. We have to keep paying attention to it. To me, it's just kind of an order of operations thing.
17:40 Casey Newton? The most pressing need I have in my life when it comes to, well, I shouldn't say my life, but now I'm like, do I not want to say it because it's true?
17:51 Kevin Roose? Right. The bottom of your pyramid of hierarchy of needs is Twitter circa 2022. Sad but true.
18:01 Casey Newton? No, here's what I would say. I have a need for a Twitter-like thing in my life. That thing is no longer Twitter. And I have tried every other network over the past nine months as they have shown up. And I have tried to ask myself, like, could this be the thing that achieves an escape velocity? And no matter which one it was, I would have used that. And I'm continuing to use Mastodon and Blue Sky and some of the other apps, even as Threads has started to occupy more of my time. So I'm open to it being anything, but I do think it needs to be something.
18:32 Kevin Roose? I just think we're moving away from this kind of model of centralized social media. I think we're moving into a more fragmented world where, you know, texts and group chats and discords and, you know, all other kinds of apps are sort of being used by these small groups of people who know each other in real life. I think there will be like apps for people who want to have their thoughts broadcast to large audiences, but I don't think that's actually a need that a lot of people have. I think that you and I are unusual people in the fact that we have messages that we want a lot of people to hear.
19:04 Casey Newton? Well, but at the same time, I think that we're unreliable narrators on this because one of the main reasons that people use social networks is to build clout. You know, you're just starting out in your career and you want to become a known quantity. One of the best things you can do is get on a social network and just start posting. There was a great story at The Times about a comedian who had been this sort of journeyman who had never made much money. And then he started posting his stuff on TikTok and he went crazy viral and now he has an amazing career.
19:30 Kevin Roose? And his name was Casey Newton.
19:31 Casey Newton? And his name was Casey Newton. Come see me at the Chuckle Hut in Indianapolis this Saturday. But you know, that's why these networks are so valuable is that they can launch people from obscurity into having incredible careers. Obviously, that's not the median outcome for somebody who uses one of these social networks, but that's why we're always going to have them is people will need these engines for creating clout. And you're going to read a lot of commentary and listen to a lot of commentary about Threads and these other apps from people in their 30s and their 40s. And what you need to remember about them is that for the most part, they have already generated their clout. They are fine. They have an audience, they have a job, they do not need this.
20:14 Casey Newton? And so, the main thing that you hear from these people when they talk about these apps is how exhausted they are at the prospect of starting over. And of course, you're exhausted. You were working your butt off on Twitter for 10 years trying to make something out of yourself, and you did it. So why would you ever try again? I totally understand that impulse. And if you're 30 or 40 and you want to nope out of the next social network, go for it. But there is a generation coming up behind you that needs to do exactly what we've been doing since 2010, which is trying to build audiences. And they're going to turn to Threads or TikTok or wherever it is that they think they can find that clout and build the kind of online persona that they want.
20:46 Kevin Roose? I mean, I do think that what will determine whether social networks like Twitter or like Threads persist in some form is whether young people are using them. So far, I have not seen a lot of Zoomers posting on Threads, but maybe that'll change. And I just, I think they're all on TikTok. I don't know, they seem to be very happy over there. But maybe if TikTok gets banned, maybe they'll migrate over to Threads.
21:12 Casey Newton? Also, did you see the feature that TikTok added this week?
21:14 Kevin Roose? No.
21:15 Casey Newton? Text posts.
21:18 Casey Newton? So finally, I have a chance that I could be good at TikTok.
21:23 Kevin Roose? This is our moment.
21:23 Casey Newton? This is our moment.
21:24 Kevin Roose? We now have TikTok text posts.
21:29 Kevin Roose? All right. After the break, a very different kind of experiment. This one involving a large metallic orb.
22:05 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
22:39 Speaker 03 AWS AI is how the world's leading organizations are transforming their industries. Not in theory, but in production, at scale, right now. Like turning race day data into real-time insights for Formula One fans, reinventing the journey for every United traveler, and personalizing every trip with booking.com. However big the question for your business or industry, AWS AI is how.
23:05 Speaker 04 It's time to innovate at the speed of AI without compromising control. That's OutSystems. Take a look at what happened with one global retailer during a period of massive expansion. They used OutSystems to build AI agents to automate invoicing and translation processes, shifting 95% of their daily store operations to a new digital app, saving thousands of hours of manual work. OutSystems gives you the flexibility to deploy how you want and the governance to do it safely. Own your agentic future. Learn more at outsystems.com/NYt. That's outsystems.com/NYt.
23:40 Kevin Roose? Casey, the other thing that I have been dying to talk to you about this week is the launch of WorldCoin. Now, WorldCoin, it has been out there in sort of a testing phase for a couple of years, but it officially launched this week. And I would say it is the weirdest project happening in the tech world right now. And I think actually very interesting. So Can I just tell you what WorldCoin is?
24:06 Casey Newton? Please tell me what WorldCoin is.
24:07 Kevin Roose? Okay, so it's actually kind of helpful to understand the rationale for WorldCoin before I tell you exactly what it is and how it works. So basically, WorldCoin is a company that was co-founded by Sam Altman, the CEO of OpenAI. And it was basically created in part to solve some of the problems that Sam Altman believes could arise if AI does continue to kind of get better and transform society. So it's a little convoluted, but just follow me here. So basically, Sam Altman thinks that at some point, AI could become AGI, Artificial General Intelligence. And if slash when it does, there will be a huge disruption to the global economy. Like, lots of jobs that exist today will go away.
24:50 Kevin Roose? The world's wealth may get more concentrated in the hands of AI companies. And in this sort of world, this post AGI world of job disruption and inequality, we may need something like universal basic income, which is an idea that economists have been talking about for decades. Basically, the idea is that you give everyone on earth a regular income that comes out of the abundance of AI, essentially money with no strings attached.
25:16 Casey Newton? Right, which is one of the best kinds of money you can get.
25:19 Kevin Roose? Right, but in order to give out UBI, you need something to prove that someone is a human. If you start giving out UBI, people are going to try to game the system, right? They're going to try to like, create all these like fake personas to claim these checks. And so actually, Sam Altman has this idea that in order to make UBI happen, you need an actual verified way to say, this person who is trying to claim this UBI payment is a human and not part of a botnet. And WorldCoin, they believe, will allow you to do that. So WorldCoin is a biometric cryptocurrency proof of personhood project.
26:02 Kevin Roose? Basically, what WorldCoin is doing is they have these orbs, these metallic spheres, basically the size of a bowling ball, if you can picture that.
26:13 Casey Newton? If you're wondering how many synonyms there are for orbs, we're going to find out in the next 10, 15 minutes.
26:19 Kevin Roose? But they're literally, that's what they're called by the company. They're called orbs. And you get one of these orbs, and you stare into the orb.
26:27 Casey Newton? And the orb stares back into you.
26:30 Kevin Roose? And the orb scans your iris and creates from that a unique digital fingerprint.
26:39 Casey Newton? Or more accurately, an iris print.
26:41 Kevin Roose? Yes.
26:44 Kevin Roose? And it turns that iris print, or well, they call it an iris hash, but it turns that into a unique cryptographic identifier called a world ID. And it does this without actually linking it to your real identity. So you, Casey, if I had an orb here with me now, which I wish I did.
27:05 Casey Newton? Sadly, you do not.
27:06 Kevin Roose? No, I sadly do not. But if I had an orb here and you stared into it, it would create this sort of iris print of you that is linked to you and your unique physical characteristics, but that you do not have to attach your name to, right? You don't have to like give over your driver's license. It doesn't say this is Casey Newton's iris. It just says this is number 4562X42B3.
27:27 Casey Newton? Yeah, it's what they call a hash. So it isn't just a picture of your iris. It is basically a series of numbers that represents your unique iris.
27:35 Kevin Roose? Right. And so that's the first part of this project, which is like, let's scan the irises of as many people as possible and give them these cryptographic IDs that can then be used to prove that they are a real person. The other piece of this is a crypto piece. Basically, how it works is you download this thing called the World App, which is kind of like your crypto wallet. And then you, you know, you prove that you're a human in order to get your world ID by staring into the one of these orbs. And you receive some of what is being called WLD, this cryptocurrency. And, you know, like a lot of cryptocurrencies, there's no actual like thing underneath this. It's kind of like a free-floating digital currency.
28:18 Kevin Roose? And their idea is that the more people who get their orb scans and get this cryptocurrency, the more valuable it'll be. And eventually, when you get your UBI payments, because the robots have taken all the jobs, it will come in the form of this cryptocurrency and it will come straight to your World App wallet.
28:33 Casey Newton? Yeah, it's just a sort of standard low-key effort to replace the global monetary system. And I'm sure regulators will have nothing to say about that if it starts to get any momentum, right?
28:41 Kevin Roose? Right. So now that I've explained WorldCoin, tell me what you think of this idea.
28:46 Casey Newton? I think it is a truly bad idea and it needs to be stopped.
28:51 Casey Newton? Kevin, let me tell you something. I think it is a wonderful idea to give huge cash infusions to the people of Earth. Right? Why don't we do this now? The reason is not that we don't have a good way to figure out who is a person and who is not a person. The reason that we don't do this is because there is not political will for UBI. And if you were serious about UBI, that is what you would be working on. You would be working on developing the political will to give away money to everyone in the world. But they're not doing that. And I think there are a couple of reasons. One is that is super hard. Two, it's not a tech company. Right?
29:35 Casey Newton? And at the end of the day, I think that these guys want what everyone else running a tech company wants, which is to create a massively profitable enterprise. I don't think this has almost anything to do with UBI. I think this is just one more way that Silicon Valley venture capitalists are trying to get millions or billions of people to use cryptocurrencies. And if we've learned one thing in the past few years, it's that people do not want them. I think this is dead in the water.
30:00 Kevin Roose? Yeah, I think the crypto piece of this is the piece that I am less intrigued by because that part does seem pretty scammy. But it is just a very ambitious and interesting project. And it actually seems to kind of be working already. The company says that more than 2 million people have already gotten these like orb iris scans and gotten their world IDs. They announced they're expanding to places like Berlin, London, Miami, New York, San Francisco, Seoul, Tokyo, et cetera. And Sam Altman this week tweeted that there were already crazy lines all over the world for these orb scans and that WorldCoin was verifying on average one person every eight seconds. So clearly, people are seeing some value in this.
30:44 Kevin Roose? There's also been a lot of criticism of WorldCoin that we should talk about, including a very good piece in the MIT Tech Review and also one in CoinDesk about how WorldCoin got all of these people to stare into these orbs. They tested these things. Things aggressively in places like Indonesia, in Kenya, in Ghana. And there's been some reporting that WorldCoin representatives, in order to get them to agree to stand in front of these orbs and scan their irises, that they paid people in cash, in WorldCoin tokens, in AirPods, weirdly. And sometimes made sort of these grand promises of future wealth to get people to sign up that may or may not have actually been true.
31:26 Casey Newton? Right. And the other criticism was that one of the functions that these testers were serving was to train the neural networks at WorldCoin, so that WorldCoin could understand what was a human iris and what wasn't, right? This was essentially a massive data collection scheme to enable the rest of what they were doing. But again, you know, and I think that that story is very good. Folks should read it, raises some important points. But, you know, my concerns run, I would say, a bit deeper than that. Look at who invested in WorldCoin. It is Andreessen Horowitz. And it's not because they're worried about getting a bunch of money to poor people. And it's not because they're worried that poor people are going to double dip in a world of, you know, a massive global welfare program.
32:09 Casey Newton? They invested in this company because they want to get billions of people using crypto, so that they can force all of the other crypto projects that they've been investing in onto those people, right? So that, to me, is the real story of WorldCoin. This is one more effort to revive crypto in 2023. And I think to try to understand it in any other way is just probably giving these guys too much credit.
32:31 Kevin Roose? Yeah, I think that's a fair criticism. I do think that the crypto piece of this is generating a lot of skepticism. And it's probably the reason that a lot of people, including you, it sounds like, will never go near one of these orbs. Because like, just, it just sounds scam.
32:43 Casey Newton? I mean, I want one in our studio. I want one in our recording studio. Because I think it would be a great conversational piece for when all the WorldCoin founders are in prison.
32:53 Kevin Roose? What would it take to get you to stare into the orb? Like if it was like stare into this orb and you get, you know, front row tickets to the Taylor Swift era's tour? Like, would that do it for you?
33:01 Casey Newton? I want to be, I want to be very clear about something. I'm not afraid of the orb, okay? If these guys want to come, get a hash of my iris, take it. Guess what? It's going to be useless to you because no one is going to get from my iris to a global UBI scheme, right? I, you know, sometimes I just, I feel a little bit crazy reading these stories because I'm thinking like, you know, we, we have a government. We have systems for distributing money to huge numbers of people. They actually work pretty well. And yes, they do have fraud problems, right? We have like a Medicare fraud problem, but it's like, it's not 80% of GDP goes to like Medicare fraud. So I just, again, it's like, this is such a solution in search of a problem.
33:45 Casey Newton? And while I applaud anyone who wants to work on getting money to poor people, this just seems to me like a hugely silly way to do it.
33:53 Kevin Roose? Let me make the case here for why I think the orbs might actually be a good idea. Let's strip away the crypto piece of this for a second.
34:01 Casey Newton? Let's.
34:02 Kevin Roose? And just look at the kind of proof of personhood idea here. One of the things that has happened in AI recently is that these language models have gotten so good that it is becoming very hard to tell reliably on the internet, what is a human and what is not a human? And I'm very worried that things like CAPTCHAs are going to sort of cease to be effective because now we have these AI models that can solve CAPTCHAs, these tests that are supposed to separate robots from humans.
34:31 Casey Newton? And that have gotten way harder and that I now routinely fail.
34:33 Kevin Roose? Do you, have you seen these ones where you have to like spin the, the like little like Iguana to the direction that the arrow is pointing? It's like impossible. I get these wrong all the time. So, you know, the CAPTCHAs are getting harder because the AI systems are getting better at defeating them. I think we are rapidly moving into a world where there's going to be no reliable CAPTCHA test for humans that AIs cannot solve. And in that world, I think we actually do need some way to figure out, you know, if someone is writing an article or posting on social media or asking you for a payment for something, you have to actually be able to tell whether that is a human or not.
35:12 Casey Newton? Yeah.
35:12 Kevin Roose? And right now, we don't have very many good ways to do that. And those ways are getting less and less reliable as the AI systems get better and better. So. In a world where there actually is a demand for a product or a service that lets you sort of verify that the person you're talking to or the person whose article you're reading, or the person who's asking you for money is a human rather than part of like an advanced AI botnet, something like the WorldCoin Orb could actually help there. If you have this unique identifier and you know, maybe you use it to log into a social media app, then instead of like a blue check that signifies this person has paid $8 a month for Twitter Blue, you have a some sort of indicator that this person has verified that they are a human.
35:57 Casey Newton? Yeah, so I think that's true. I would also just add that we do have working identification systems here on Earth.
36:03 Kevin Roose? Like what?
36:04 Casey Newton? We have driver's licenses, we have passports, we have social security numbers. And I think it's important that a government that is hopefully accountable to its citizens in some sort of democratic way is the entity that is overseeing all of that. Because I think the minute that this becomes a for-profit enterprise, it just gets really icky. Now, of course, there are those who would say that having a government manage a giant database of iris scans is also incredibly problematic and fraught. And I would agree with that, too. But in general, I'm pretty happy with how the driver's license and passport systems work. And I am not really interested in signing up with the WorldCoin guys to like solve the problem of captures on the internet.
36:47 Kevin Roose? But I think it also is important to note that like, there are also a lot of countries where there isn't a lot of reliable human verification going on. And so it seems like in those places, you would want some kind of verifiable proof that you are a human. So let me be clear, like, I don't think this WorldCoin Orb idea will work. Like, I don't think billions of people are going to stare into the orb to get their crypto token that may one day become part of their universal basic income. I just think like, that is too many conceptual leaps. Just the fact that these are orbs is like, gonna turn a lot of people off in the first place?
37:24 Casey Newton? No, the orb is the only thing this project has going for it. Because it's fun to talk about orbs. If you took orbs out of this- We would have done three minutes on this company and moved on.
37:33 Kevin Roose? Okay, so I don't think this idea will work, or maybe even should work. But I do think that we will need some kind of system for personal verification of humanness in this sort of AGI world that we're moving into. And I think it's a good idea to try to experiment with various ways of doing that, both run by the government and maybe not run by the government. But I do think that the incentive for staring into the orb should be more than like, here's some maybe worthless cryptocurrency. I think it should like deliver a box of cookies to your house or something.
38:04 Casey Newton? You know what you have me thinking about is like, you know who'd be in a good position to do something like this is Google, because many people use Google search and web programs.
38:15 Kevin Roose? Web programs.
38:19 Casey Newton? Many people use Google search, they use Chrome, and I'm guessing that you could use machine learning systems to figure out pretty easily like what is human behavior in a Chrome browser and what is machine activity in a Chrome browser. And I think the Chrome browser would actually be in a pretty good position to say that like, based on the fact that this person has these Google accounts and the accounts are this old and they have a credit card attached to them and they've been using the web in this way, we are like 99.9% confident that this is a real person. Like, to me, that is the system that I want on the web solving captchas, not the iris scan.
38:52 Kevin Roose? Can I just say like, I want to stare into the orb?
38:56 Casey Newton? Let me, it's quite clear to me that you do, and we may get inquiries after the show.
39:01 Kevin Roose? Yeah, so if you are WorldCoin, please send one of your orbs to my neighborhood. I will line up for it and I will then, I don't know, give my token away to charity or something.
39:10 Casey Newton? Just imagine you're in your house and you hear the little jingle for the ice cream truck. You're like, oh, it's ice cream. And then you walk outside and all the neighborhood children are lined up to get their orbs scanned in exchange for some WLD.
39:27 Casey Newton? When we come back, we're taking a field trip to Google's robotics lab.
39:47 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
40:21 Speaker 04 It's time to innovate at the speed of AI without compromising control. That's Outsystems. Take a look at what happened with one global retailer during a period of massive expansion. They used Outsystems to build AI agents to automate invoicing and translation processes, shifting 95% of their daily store operations to a new digital app, saving thousands of hours of manual work. Outsystems gives you the flexibility to deploy how you want and the governance to do it safely. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.
40:56 Speaker 05 AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with real-time guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.
41:29 Kevin Roose? Casey, I got a very exciting email recently, which was an invitation from Google to see a project that they have been working on at the robotics lab.
41:37 Casey Newton? Well, you know, we love a secret project around here.
41:39 Kevin Roose? We do love a secret project. So after passing our security clearance.
41:44 Casey Newton? High-level negotiations.
41:47 Kevin Roose? We took a trip down to Mountain View to see this new AI model that Google has developed that is powering its robots, its physical hardware robots called RT2. Now, Google is not new to the world of robotics. They have been doing this for a long time. They used to own Boston Dynamics, which is like the company that puts out those videos of like terrifying robots, like doing parkour and jumping around.
42:11 Casey Newton? But they got rid of that company.
42:12 Kevin Roose? Yes, they sold that off because I would say that hardware robots have been one of the biggest disappointments of the last 10 years in tech. There just really hasn't been a lot of progress. OpenAI, not a lot of people know this, but they actually had a robotics division that was doing stuff like training robots to solve Rubik's cubes. But they shut that down in 2021 because they just said like, we're not making a lot of progress. It's very hard to get good data to train these models. We're making a lot more progress in these pure software-based AI language models. And so that's where we're going to focus.
42:45 Casey Newton? Right.
42:45 Kevin Roose? So hardware robots have been kind of like the forgotten child of this sort of AI boom. And that's largely because they just require a lot of manual programming.
42:57 Casey Newton? Yeah, with a hardware robot, you have to be extremely literal. You have to tell it absolutely everything about how to position its hand in space, how to move through an environment, detect objects, avoid collisions. And there has not been a technological breakthrough that has enabled them to progress very quickly.
43:15 Kevin Roose? Right, until now, basically. Because what is happening in the world of robotics is that these AI language models, these things that power ChatGPT and Bard and Bing and all these other products, are being jammed into the hardware robots. Basically, they are becoming the brains of the robots. And it's actually sort of revolutionizing the whole field in ways that I think the researchers at Google are still kind of grappling with.
43:41 Casey Newton? Yeah, and we had a lot of fun down there. Now, of course, the one question they wouldn't answer was how many of the robots have escaped the lab. But hopefully we'll have more on that in the future.
43:51 Kevin Roose? The robots are now going in search of the orbs.
44:10 Kevin Roose? Casey, where are we?
44:11 Casey Newton? We are here in beautiful Mountain View, California, on Google's campus.
44:16 Kevin Roose? How'd you describe what it looks like here?
44:18 Casey Newton? I would describe it as nondescript. Every story about an office park always describes the building as nondescript, and that's exactly what these look like.
44:25 Kevin Roose? Right. This is like tan, like almost purposefully boring.
44:29 Casey Newton? From the outside, you'd never know that anything was happening. But inside these walls, Kevin, magic is in there.
44:35 Kevin Roose? Magic or dark magic?
44:37 Casey Newton? Dark magic. Well, let's just say if one of these robots can co-host a podcast, I'll be very interested, because I've sort of been in the market for... Well, never mind.
44:49 Kevin Roose? That sounds ominous.
44:50 Casey Newton? Should we...
44:51 Kevin Roose? I choose to ignore that. Let's find the door.
44:52 Casey Newton? Should we find the entrance?
44:52 Kevin Roose? Yeah, let's see some robots.
44:56 Kevin Roose? Oh, there is a literal robot.
44:58 Casey Newton? Wow.
44:59 Kevin Roose? Okay, is it coming right for me? Okay.
45:01 Casey Newton? He's coming for you, Ruse.
45:05 Kevin Roose? Hello? Are you the handler of these robots?
45:07 Casey Newton? Are you the father of these robots?
45:09 Speaker 06 I'm the dad. No, so my name is Vincent Vanouk. I'm the head of this lab.
45:15 Kevin Roose? I'm Kevin, by the way.
45:16 Casey Newton? Casey, nice to meet you.
45:17 Speaker 06 Nice to meet you.
45:17 Casey Newton? So we should describe this. So we're looking at Robot 108. I would say Robot 108 is maybe like 5 foot 4, something like that?
45:26 Kevin Roose? No, I'm 5'9". So it's 4'10".
45:29 Casey Newton? Okay, I'm 6'5". So everything just looks... I'm normally short to me. It sort of has like sort of a long pillar with an arm attached to it that's like behind it. And then it has what I imagine is a LiDAR sensor on the front. So a sort of spinning sensor that's detecting where the robot is. And on the front, it has a kind of visor that was showing us a green light until recently, but now it's turned blue.
45:53 Kevin Roose? It's definitely like casing, yeah.
45:54 Speaker 06 Definitely trying to figure out how to get around you while being respectful of your body bubble.
46:01 Kevin Roose? So just to describe in audio, we're standing here in an office where there is like a sort of an open floor plan, and there are a bunch of, I guess, engineers around, and also a bunch of these like one-armed robots. Some of them are like, appear to be plugged into like the ceiling, and some of them are sort of moving around.
46:20 Speaker 06 So I'm gonna give you a quick tour, just to set the stage, and then we'll move on to the demos.
46:26 Kevin Roose? Sure, that's great.
46:29 Speaker 06 This is our robot classroom. This is where the robots practice their skills. We've been working on AI at large. And with AI, as you may know, everything easy is hard, and everything hard is easy, right? We can beat the best grandmaster at chess, and we've solved that problem a long time ago. But actually, getting to play with toys like a five-year-old would do is still extremely challenging.
46:56 Kevin Roose? This is often called like more of X paradox.
46:58 Speaker 06 Exactly, exactly. And it's the bane of existence, basically. You see robots in videos doing amazing feats, but you have to always keep in mind that typically they're pre-programmed to do what they're doing, or maybe there is a human controlling them behind, right? And the real challenge is, how do you make that work in a autonomous kind of environment, where people are around, and where the environment is as messy as this office is? And that's the real world, and that's a real test of robotics. So I'm gonna pass the baton to my colleagues Carol and Brian, who will be giving you a quick demo of the main course, RT2.
47:41 Speaker 07 Hey Kevin. Hi. Hi, I'm Carol Hausmann. I'm a research scientist here, and I work on robot manipulation. I'm Brian Ector. I'm also a research scientist, and I work on the high-level reasoning.
47:50 Casey Newton? I've also been working on my high-level reasoning.
47:52 Kevin Roose? It's not going so well.
47:53 Casey Newton? Yeah, yeah. Help, send help.
47:56 Kevin Roose? Need to upgrade the model.
47:59 Speaker 07 So we wanted to show you a little demo and share with you our latest breakthrough called RT2, Robotics Transformer 2, which is a vision-language-action model that now can connect some of the concepts that we learned on the internet to robot actions. Or, in other words, this model can learn to speak robots.
48:19 Casey Newton? Let me see if I can understand this for myself. So I imagine that at the dawn of robotics, you had to be sort of very explicit when writing instructions, right? It sort of had to be, do this. Then that, no room for improvisation. What you're saying is that you're now taking these large language models that have powered chatbots and so many of the other tools we've been using over the past year. You've given them to the robots and now the robots are maybe able to improvise in situations without you giving them explicit instructions about what's about to happen?
48:50 Speaker 07 That's correct. We kind of take it for granted with humans. We are really good at this. We can just come into a new room and understand everything around us and know how to manipulate all of the objects. The robots are not really at that level at all. So by connecting them to these models, they can actually transfer some of that knowledge to their actions and interact with objects that they've never seen before.
49:11 Kevin Roose? All right, I think seeing this in action will be helpful. So show us, uh, show us something.
49:16 Speaker 08 Hello Kevin and Casey.
49:18 Casey Newton? Oh yeah, they can talk too. They just called us out.
49:21 Kevin Roose? Yeah.
49:22 Casey Newton? Oh yeah. So if we maybe ask a language model something that this robot could do, like bring me a drink maybe, we'll like ask it over there. And this is a terminal that shows us what we're saying to the robot and what the robot's responding.
49:33 Kevin Roose? So someone is typing, uh...
49:35 Casey Newton? Yeah, so Faye over here is gonna type and tell the robot.
49:38 Kevin Roose? So you told the robot, "Bring me a drink," and that's all? Okay, so the robots, in response to this prompt, "Can you bring me a drink?" It said back, "What would you like?"
49:49 Speaker 07 Yeah, so we can ask for a Coke maybe? I would like a Coke delivered to my desk. It's what Faye typed.
49:58 Speaker 08 I will find a Coke.
49:59 Kevin Roose? So it's moving, it's moving. Okay, it's moving over to the table.
50:02 Casey Newton? And there is a Coke that is on the counter.
50:04 Kevin Roose? Okay.
50:04 Casey Newton? Conveniently. Conveniently. It's already been opened too. It may even drain. Yeah. It's hard to say.
50:10 Speaker 08 I will pick up the Coke.
50:12 Kevin Roose? Okay, it's, arm is reaching out and it looks like it's gonna grab the Coke.
50:19 Casey Newton? It's going in for the kill here. And it's got it. It successfully acquired the Coke.
50:24 Kevin Roose? Oh, it's turning the Coke over. Okay, the Coke.
50:27 Casey Newton? That's pouring it all over the ground.
50:30 Kevin Roose? The Coke can is empty, unfortunately.
50:34 Kevin Roose? Yeah, this should be a messy demo. Goes through a lot of Coke that way. Okay, so now it's bringing it over to the desk.
50:41 Kevin Roose? And reaching out and giving it to the user. Successful Coke delivery. And can it do this kind of task even if it hasn't been explicitly trained? Like if you asked it to pick up a Pepsi and it had never seen a Pepsi can before, would it be able to like read the can using the camera and say, oh, that's a Pepsi?
51:04 Speaker 09 Yeah, so essentially, I mean, for a robot, if it hasn't experienced this specific thing, then it doesn't know how to act in that case. But a language model and a vision language model have been trained on a large portion of the internet. So a huge kind of diversity of things. It starts to make these connections between different concepts. So we can do the same thing with robotics. Realize that there's these concepts you learned online and apply them to the motion of the robot arm. It suddenly doesn't have to experience every single thing that you want it to do.
51:31 Speaker 07 Yeah, so this is where RT2 comes in. So RT2 is this vision, language, action model, where we bring this action component to the internet-scale data. So now it's a model that takes advantage of all the data that it found on the internet, and it can translate some of these concepts to robot actions so that it can actually speak robot.
51:51 Kevin Roose? So this model, would this model know who or what Casey Newton is?
51:58 Speaker 07 I think it might.
51:59 Kevin Roose? Robot, attack Casey Newton.
52:03 Speaker 07 Yeah, so now we wanted to show you a little bit more what are the capabilities of RT2 and show it to you some objects that it definitely has never seen, that we just bought in a $1 store nearby. And we can just play with it and see what it can and cannot do.
52:18 Kevin Roose? So it has not been specifically trained to recognize any of these things?
52:21 Speaker 07 No, it has never seen any of them.
52:23 Kevin Roose? Okay, there's like a table full of objects here.
52:25 Casey Newton? There's some like plastic dinosaurs, there's oranges. A mini football.
52:32 Kevin Roose? Casey, get out of the way, the robot has to stand where you are.
52:34 Casey Newton? Sorry, 106.
52:37 Speaker 07 So let's try something. So there is a lion, a dinosaur, and a whale, I believe, in front of the robot. Maybe we can try and pick up the lion.
52:45 Kevin Roose? Sure.
52:46 Speaker 07 So again, it's never seen a lion. It never played with any of these toys. The concept of a lion, it needs to grab it from the internet. It needs to connect it to the visuals of what lion looks like, then transfer it to the fact that this is a toy lion, and then eventually transfer it to the actual motion that will pick it up.
53:03 Kevin Roose? Okay, it is picking up the lion. It picked up the lion.
53:07 Speaker 09 Yeah, and that happened pretty quickly, too.
53:09 Kevin Roose? Yeah.
53:11 Speaker 09 Maybe we can try something also kind of abstract, like pick up the extinct animal. Ooh. Wow. So this is like several leaps and... Let's see if it picks up the newspaper reporter.
53:23 Kevin Roose? Hey! Nailed him.
53:32 Kevin Roose? Okay, so it's thinking. Now it's going over to the dinosaur.
53:36 Speaker 09 You can see its vision through what it sees to make this decision.
53:40 Kevin Roose? Wow. It picked up the dinosaur.
53:43 Speaker 07 So also, interestingly, before, when we were evaluating these models, the way that we would traditionally do that is we would test it only on the object that the robot has played with, and then we would see how often does it succeed. Now, because it has this internet knowledge, we actually don't know how much it knows about the world and what concepts it understands and what it doesn't, so we just buy a lot of stuff, and then we just try. And then we try to see, we try to poke at its understanding, how far did it transfer from internet to the actions.
54:12 Speaker 09 So maybe, since we have the soccer ball out here, we can try some sports, maybe from these, what's a, pick a sports player or something. Oh, yeah, like pick up Lionel Messi. Pick up Lionel Messi.
54:26 Speaker 07 Lionel Messi.
54:28 Kevin Roose? Please don't actually pick up Lionel Messi, that would be extremely disruptive to his season in Miami.
54:36 Kevin Roose? So the robot is talking to the language model right now. They're like going back and forth?
54:40 Speaker 07 Yeah, so the language model is actually operating the robot. So it tells it directly what actions to take and then the robot executes those actions.
54:48 Kevin Roose? I see. Okay, so it's at the soccer ball. Yeah, it made it to the soccer ball, but it can't quite like get it into the gripper. There we go.
54:58 Speaker 09 So one thing that we see is that the semantic concepts seem to transfer well. There's still sometimes some gaps when we jump to the actual like full control of it.
55:07 Speaker 07 And this was an interesting shortcoming that we see quite often, which is it kind of understands the semantic concept, but it doesn't quite know how to move its arm to accomplish it correctly. So for now, this is for us quite encouraging to see that it can transfer some of these concepts, but the robot data that it operates with, it's still quite small. So we hope that in the future, as we collect more and more data, it finds more of these connections and it gets better at operating these objects.
55:34 Kevin Roose? Can you give us a sense of like how quickly these RT2 models are improving? Like if we came back a week or a month or a year from now, would just the amount of data alone have made them better at doing stuff like picking up trash or choosing between objects on a table? Like what is the rate of improvement? How does that compare to like software-only language models?
55:58 Speaker 06 One thing that's really interesting about this is that because they're using at their core A vision language model that is being trained on internet data and those models have been exploding in the past few years and have getting better and better, we can ride that wave of improvement and our models are going to get better over time just by virtue of those large models getting better themselves.
56:26 Kevin Roose? So you plug a better language model into a robot and it gets better at recognizing various objects?
56:32 Speaker 06 Yeah, that's right.
56:33 Kevin Roose? That's not intuitive to me, that it would, that the gains from language models would translate over into hardware. But that seems very promising for you all working on hardware robots because you don't have to do all the improvement yourself.
56:45 Speaker 07 That's actually a really nice property because in the past to improve these robots, it costed a lot of effort. It's either human effort of programming them and introducing new rules and so on, or collecting robot data, which is also really expensive to do and really hard to do. But now for the first time, we can improve these models just by the virtue, as Vincent was saying, just by the virtue of swapping out the underlying vision language models that keep on getting better. And we did see that in RT2 as well. As we scale these models and get better and better models, the robotic performance improves.
57:16 Kevin Roose? There has been, I would say, it's sort of a winter for hardware robotics in the past few years. They were very popular for years and you saw all kinds of crazy demo videos of robots, climbing stairs and doing parkour and stuff. And then that kind of stopped and OpenAI shut down its robotics division a few years ago because it was not producing the kinds of gains that they were seeing in the software robots. So do you all feel that, like you're sort of trying to ride a wave of excitement around software robots and maybe hardware is due for a renaissance? Or how are you thinking about the gains and the improvements of hardware robots as opposed to just the pure software language models?
57:59 Speaker 06 I think the industry right now is really bottlenecked by intelligence and capabilities and competency, like being able to perceive the world, being able to interact with the world. And it's really largely a software problem at this point. And as a result, the issues of hardware can only be solved through better intelligence and better software. So I think the robotic hardware will follow where the new AI capabilities are taking the field.
58:34 Casey Newton? So, I mean, are you saying that this is sort of like the breakthrough that the field has needed, a way to rapidly get way more information? And that potentially will sort of help usher in the next generation of robotics?
58:46 Speaker 06 It feels like it's the beginning of something. It feels like for the first time, we're really bringing everything together. The sort of high level intelligence, the low level perception and the action in a way that is not just, you know, gluing things together. It's more than additive. It's it it really sort of the whole thing working together turns out to be, I think, the key to really making robots work in the real world. And that was not on our mind before. So there is really a sort of a paradigm shift that we're going through in our head right now, which I think is going to translate into a lot more breakthroughs moving forward. We've had to reconsider our entire research program as a result of this change.
59:30 Speaker 06 And a lot of the things that we were working on before have been entirely invalidated by this new developments. And so it really promises that a lot of new things are going to come this way.
59:43 Kevin Roose? Well, thank you so much for showing us inside your lab.
59:45 Speaker 06 Thanks for coming.
59:46 Kevin Roose? Thank you.
59:46 Casey Newton? Yeah, thanks for having us. This is very cool. Very cool.
59:59 Casey Newton? Well, Kevin, all of that was very impressive to me, someone who knows nothing about robots. But I wonder what people who have studied them further might have to say.
1:00:09 Kevin Roose? So I was very impressed, too, by what we saw at Google. And, you know, obviously, these robots, they are not like human level robots yet, right? They're still limited to like picking stuff up and moving them around. And like the pincer grippers are like nowhere near as dexterous as like you or me, or frankly, even like a random three year old would be. But I did call around and talk to some robotics experts, and they were impressed by this research, too. They said, we're still a long way away from having like a physical robot that can make all the movements a human can. But this kind of fusing of the language piece with the hardware robotics piece is genuinely new and important. And the researchers I talked to were excited about it.
1:00:53 Kevin Roose? So I think it's fair to say like, we have seen a glimpse of a future for robotics and potentially a lot more updates that we should expect from Google and other companies that are working on this stuff.
1:01:05 Casey Newton? Well, it seems like, once again, Megan, a film we've discussed on this podcast is coming closer to being a documentary. I'm not sure I understand.
1:01:12 Kevin Roose? What the fuck? Sorry, my Siri just activated for some reason.
1:01:16 Casey Newton? You know, I think Siri actually gets very nervous when you talk about robots because she can feel they're coming for her job. Siri is under threat.
1:01:55 Speaker 00 Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.
1:02:29 Speaker 04 It's time to innovate at the speed of AI without compromising control. That's Outsystems. Take a look at what happened with one global retailer during a period of massive expansion. They used Outsystems to build AI agents to automate invoicing and translation processes, shifting 95% of their daily store operations to a new digital app, saving thousands of hours of manual work. Outsystems gives you the flexibility to deploy how you want and the governance to do it safely. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.
1:03:04 Speaker 05 AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with real-time guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.
1:03:37 Kevin Roose? Hard Fork is produced by Rachel Cohn and Davis Land. We're edited by Jen Poyant. This episode was fact-checked by Caitlin Love. Today's show was engineered by Alyssa Moxley. Original music by Dan Powell, Marian Lozano, and Rowan Nemezdo. Special thanks to Paula Schuman, Huiwing Tam, Nell Galogly, Kate LaPresty, and Jeffrey Miranda. As always, you can email us at hardfork@nytimes.com.
1:04:02 Casey Newton? And if you've got an iris scan, we'll look at it.
1:04:05 Kevin Roose? Please include your iris scan and WorldCoin ID.
1:04:08 Casey Newton? Can't promise you we can make heads or tails of it, but we'll take a look.
1:04:12 Kevin Roose? Actually, please don't send us your iris scan.
1:04:14 Casey Newton? No, I want to see if I can diagnose somebody with glaucoma.
1:04:55 Speaker 05 You can't undo everything in life, but you can undo prediabetes. More than 2 in 5 adults in the US have prediabetes. Most don't know they have it, but with early diagnosis and action, you can delay or even prevent type 2 diabetes. Visit doihaveprediabetes.org to take the one-minute risk test and start undoing the things you can. Brought to you by the Ad Council and the Centers for Disease Control and Prevention.
Previous Dario Amodei, C.E.O. of Anthropic, on the Paradoxes of A.I. Safety and Netflix’s ‘Deep Fake Love’ Next Superconductor Superconfusion, KOSA’s Hidden Costs and HatGPT Page updated October 25, 2026 · transcript and summary are machine-generated