Latest / The Nation Podcasts / What’s Driving the Push For Humanoid Robots w/ James Vincent | Tech Won't Save Us

Transcript

What’s Driving the Push For Humanoid Robots w/ James Vincent | Tech Won't Save Us · The Nation Podcasts

0:00
[SPEAKER_01]: I think we're very susceptible to being manipulated for the same reason that I've talked about over exaggerations of capability.
0:06
[SPEAKER_01]: We see a human form and we start treating it like a human.
0:09
[SPEAKER_01]: I think that's very dangerous and it's something that's going to be a real conversation that we're really going to have to consider seriously over the coming decades.
0:33
[SPEAKER_00]: Hello and welcome to Tech Won't Save Us Made in Partnership with the Nation magazine.
0:36
[SPEAKER_00]: I'm your host Paris Marks and this week my guest is James Vincent.
0:40
[SPEAKER_00]: James is a journalist and writer based in the UK and author of Beyond Measure, the Hidden History of Measurement.
0:46
[SPEAKER_00]: He's also a former senior editor at The Verge.
0:49
[SPEAKER_00]: Now, I'm sure that you have seen all of these videos circulating online for the past number of months, the past year of humanoid robots dancing and trying to do housework and doing various other tasks, right?
1:04
[SPEAKER_00]: Because there are so many companies that are now working on these humanoid robots, certainly inspired by Elon Musk and what Tesla has been doing with Optimus.
1:14
[SPEAKER_00]: who are trying to create a market for these things, right?
1:17
[SPEAKER_00]: And who are trying to convince us that humanoid robots are the future that they're going to be a really important part of our lives, or maybe even going to take over a lot of jobs, but to be honest, that narrative seems a bit far-fetched to me, and probably to you too, if you are a listener of this show.
1:36
[SPEAKER_00]: And so James recently wrote a feature piece for Harper's where he really dug into what is behind these humanoid robots and where they actually are on the curve of innovation or whatever.
1:48
[SPEAKER_00]: You know, are these robots a technology that are really about to take off and change everything?
1:54
[SPEAKER_00]: Or is this just like a load of hype that is driving investment in the certain companies, but is really not going to pay off?
2:02
[SPEAKER_00]: in the way that these CEOs and executives and what have you want us to believe.
2:08
[SPEAKER_00]: James is really dug deep on this technology, and I was really excited to hear his perspective on where this all is, how connected it is to the janitor of AI moment, and what we're seeing in the difference between, say, the United States and China in the approach to developing these robots,
2:23
[SPEAKER_00]: And whether it's looking like one of those countries is poised to do a lot better than the other.
2:28
[SPEAKER_00]: So with that said, I think you're really going to enjoy this conversation.
2:31
[SPEAKER_00]: Hopefully it gives you insights into what is going on with all of these robots and all these videos that you're seeing in your social media feeds.
2:38
[SPEAKER_00]: So what that said, if you do enjoy this episode, make sure to leave the five star of you and your podcast platform of choice, you can share the show on social media or with any friends or colleagues who you think would learn from it.
2:47
[SPEAKER_00]: And if you do want to support the work that goes into making tech won't save us every single week.
2:51
[SPEAKER_00]: So we can keep having these critical in-depth conversations that inform you about what's happening and so many different parts of the tech industry.
2:58
[SPEAKER_00]: You can join supporters like Vicki from Ottawa, Wayne and Winnipeg and Wade from Indiana,
3:04
[SPEAKER_00]: by going to patreon.com slash tech won't save us, where you can become a supporter as well.
3:09
[SPEAKER_00]: Thanks so much and enjoy this week's conversation.
3:12
[SPEAKER_00]: James, welcome to tech won't save us.
3:14
[SPEAKER_00]: Alright, thank you for having me.
3:15
[SPEAKER_00]: Really excited to have you on the show.
3:17
[SPEAKER_00]: It's been far too long.
3:18
[SPEAKER_00]: I have to admit, I meant to read your book beyond measure when it came out and it was just one of those ones that was like on the to read list and the to read list got so long and it's still on there and hopefully I will get to it.
3:30
[SPEAKER_01]: It's still available in good bookstores.
3:32
[SPEAKER_01]: I would say, you know, if you're a nerdy person, it's the history of measurement.
3:35
[SPEAKER_01]: I'll do a quick pitch for it.
3:36
[SPEAKER_01]: It's, it's still all relevant, you know, the history of the cube.
3:38
[SPEAKER_01]: It hasn't changed in a few thousand years, so it's all there to read.
3:42
[SPEAKER_00]: We haven't developed a whole new measurement system that we're using since it was published.
3:45
[SPEAKER_01]: I mean, we have outside of the US and, you know, Canada, sorry, Canada is, is matrix.
3:51
[SPEAKER_01]: Sorry.
3:52
[SPEAKER_01]: We're like mostly, man.
3:54
[SPEAKER_00]: It's a yeah, you'll let the UK on you.
3:57
[SPEAKER_00]: You have a little bit of both exactly, but in different ways like I think you guys still use miles on the on the road But we we use kilometers, but when we when we when we cook we have Fahrenheit on our like ovens and stoves and whatnot So yeah, yeah, oh, no, we we've got Fahrenheit well actually.
4:11
[SPEAKER_01]: I think we've got dual there Celsius and Fahrenheit depends.
4:14
[SPEAKER_01]: They're all the instructions have both Anyway, fascinating subject, which all your listeners can read more about
4:20
[SPEAKER_00]: I'm fascinated by measurement and that kind of stuff so yeah, I do need to get to it people should check it out But anyway, we're talking about something completely different.
4:29
[SPEAKER_00]: You know, we have been hearing a ton about these humanoid robots lately And you have this like fantastic story in Harper's really digging into the state of all of that Why was that a subject that you were like you know what I want to spend some time really digging into this exploring it further understanding what's going on here
4:47
[SPEAKER_01]: I mean, I think just because robots, humanoid robots in particular, that's what the focus is of the story we're talking about robots that are bipedal, that have arms, that mimic human morphology, they were everywhere last year.
4:58
[SPEAKER_01]: You were seeing them running on race track, you were seeing people fighting them in rings, you were seeing them doing backflip, people hitting them with chairs, looks very fun, I wanted to try it for myself.
5:07
[SPEAKER_01]: And, you know, that led me down the road of like, okay, what is the actual stakes here?
5:12
[SPEAKER_01]: Because I think in many ways, the tech industry, you know, obviously, you know, it's very intimately, has divorced itself better and further for reality over the last few years.
5:21
[SPEAKER_01]: We've had so many big projects and products, you know, from NFT to virtual reality, that have made these past claims on our attention and on investors money and said, look, we're going to deliver the future and they've not come off.
5:33
[SPEAKER_01]: And I've been following robots
5:35
[SPEAKER_01]: for ages.
5:36
[SPEAKER_01]: So years and years and years as a as a reporter at the Virgin doing stories elsewhere.
5:40
[SPEAKER_01]: And in the last year I thought this was when they reached their sort of tipping point.
5:46
[SPEAKER_01]: The phrase within the industry which was used is that robots were having their chat GPT moment, you know, that their capabilities were about to explode just in the way that large language models had and that there would be a correlation in terms of revenue and investment and all of this sort of
6:01
[SPEAKER_01]: You know, the stage we're at now and I think 2026 is going to be a bit of a reckoning.
6:06
[SPEAKER_01]: And we're going to see actually whether there is a man behind the curtain pulling the strings on all these things.
6:11
[SPEAKER_00]: Yeah, I wonder if the reckoning is coming for AI as well.
6:14
[SPEAKER_00]: But I'm happy you admitted that the whole reason you did this was because you wanted to hit a robot with a chair.
6:21
[SPEAKER_01]: Yeah, I mean, so that's the name of the piece that I wrote kicking robots and I did, I unfortunately didn't get to kick on.
6:28
[SPEAKER_01]: I always told that that would be a little bit too dangerous, but I went to the headquarters of, you know, one of the leading U.S. firms, which is called Appronic, and I got to shove one of their robots about with a big stick, and that was very satisfying to me, and also incredibly uncanny.
6:42
[SPEAKER_01]: You know, I have to say we're going to be talking a lot about height and where these machines fail to meet expectations.
6:48
[SPEAKER_01]: But there is something about when you first come face-to-face with a humanoid robot, stands as tall as you, weighs about the same as you.
6:55
[SPEAKER_01]: And I remember giving this thing a push, just a light push at first, because I was like, I don't want to break anything.
6:58
[SPEAKER_01]: And it just, it was so solid.
7:00
[SPEAKER_01]: It was rock solid, and I started shoving it hard and hard.
7:02
[SPEAKER_01]: I gave it a big shove with a stick, and it sort of swung backwards in the air.
7:06
[SPEAKER_01]: And it swung it so arms out towards me.
7:09
[SPEAKER_01]: It was just such a completely human motion of someone arresting themselves mid air, adjusting their sort of inertia, their momentum.
7:16
[SPEAKER_01]: And yeah, it just amazed me entirely, and that sort of, you know, it made it my fascination for that to see how real these things really are.
7:25
[SPEAKER_00]: I have to imagine that standing face to face with one of those robots, it can be a bit eerie as well, right, to have this thing that like is in the form of a human.
7:34
[SPEAKER_00]: Obviously, it doesn't have like the skin and stuff that we have, but and also like,
7:38
[SPEAKER_00]: You know, it is reacting in some sense, right?
7:40
[SPEAKER_00]: Like it makes me think of the conversations that we were having about, Chanti BT in the past couple of years, where it's like, it feels like you're talking to something.
7:47
[SPEAKER_00]: It feels like the way that it's presented that it, there must be some like consciousness, stop behind it.
7:53
[SPEAKER_00]: I imagine running into one of these robots, like, it's not the same feeling as that.
7:57
[SPEAKER_00]: You're not talking to it, but there's, there must be a different kind of feeling of like, oh, this feels like it's something and it can maybe lead you to also see it as more than what it is as well.
8:07
[SPEAKER_01]: Absolutely.
8:08
[SPEAKER_01]: I think this is, you know, why it seems too obvious, but this is the crux of the entire industry that the people look like people, you know, some people might hear that and think, yeah, silly, but, you know, we got to get to the underlying technology underneath and we got to talk about the harder and the speeds and other stuff.
8:24
[SPEAKER_01]: That's all very important as well.
8:26
[SPEAKER_01]: But there is something
8:28
[SPEAKER_01]: as you say, uncanny, eerie, possibly awe-inspiring about the fact that these look like people, and this creates a reaction that actually moves beyond rationality sometimes.
8:40
[SPEAKER_01]: You know, there are sort of explicable ways to think about it.
8:43
[SPEAKER_01]: One of my favorite skeptics on robotics is a guy named Rodney Brooks, who is, you know, a very storied robot assist himself.
8:50
[SPEAKER_01]: He did rethink robotics.
8:51
[SPEAKER_01]: He did the Sawyer robot, all of this sort of stuff.
8:53
[SPEAKER_01]: And he, you know, makes this very useful, enlightening observation, which is just that we see a form, and then we imagine the capability based on that form.
9:01
[SPEAKER_01]: We see something that looks like a human, and we just assume it can do all the things that a human can do in the same way that before CHAPGPT and their ilk came about.
9:11
[SPEAKER_01]: If we interacted with anything with that level of fluency and speech, that ability to hold a conversation to go back and forth, we assume this cognition behind that skill, because that's the only time we've ever encountered speech as a mark of cognition of a thinking being.
9:25
[SPEAKER_01]: So when we see a robot walking about and doing these tasks, we immediately go, it's a human, shake like a human.
9:32
[SPEAKER_01]: It must do all the human things we can do.
9:33
[SPEAKER_01]: And this leaves us to such a over exaggeration of capabilities.
9:38
[SPEAKER_01]: So that's one aspect of it.
9:40
[SPEAKER_01]: And this economic is that the VC, the tech world is so frothy and has been for many years at the moment that that is exactly the sort of quote-unquote magic that will get you billions of dollars invested in your company.
9:52
[SPEAKER_01]: And the fact of the matter is that humanoids, the hardware has advanced hugely over the past five, six years, something like this.
9:58
[SPEAKER_01]: It's got cheaper.
10:00
[SPEAKER_01]: This has been downstream of various innovations we've seen in the EV industry, which has made battery tech better.
10:05
[SPEAKER_01]: We've got better electric actuators, we've got cheaper sensors, all of this stuff.
10:09
[SPEAKER_01]: All of this means is that that actually, it incredibly easy to put together a bog-standard humanoid that 20 years ago would have earned billions of dollars of funding from DARPA, right?
10:19
[SPEAKER_01]: It would have been world-breaking
10:21
[SPEAKER_01]: And now, you can get to that point with a few million dollars.
10:25
[SPEAKER_01]: And that means that if you have the show, then you can take that on the road.
10:29
[SPEAKER_01]: And this is where a lot of the excitement and I think the bubble in human oil is coming from in the moment.
10:35
[SPEAKER_00]: So I think that sets us up really well, right, to explore what is going on.
10:40
[SPEAKER_00]: in this moment because you know you mentioned that we're seeing a lot of videos now of these humanoid robots and have been over the past year or two and even before that we would regularly see these videos of like the Boston Dynamics robots right originally the dog and then moving up to the more kind of like humanoid figure that they had as well and as you say you'd see them try to push it over and it would it would come back and stuff but it feels like in the past
11:04
[SPEAKER_00]: few years we have had this like significant proliferation of the number of companies who seem to be trying to do the same thing.
11:13
[SPEAKER_00]: And so I guess I wonder why is it that we're seeing so many companies go into this space right now and why are we seeing so many tech executives talk about, you know, the humanoid robots being on the cusp of like mass usage, mass consumption rolling out everywhere.
11:28
[SPEAKER_00]: Why is it this moment that this is all happening?
11:30
[SPEAKER_01]: So the barriers of entry have dropped, as I mentioned, the hardware is cheaper, it's easier to get a seat at the table to build a basic robot, and then it's the chat GPT moment.
11:39
[SPEAKER_01]: Basically, the current thesis for human-ord companies that we're seeing in the US and China, particularly, and a little bit in Europe as well, is that the same sort of technological transformation that took place within deep learning can now be applied to robotics.
11:55
[SPEAKER_01]: So the two paradigms that work fair are end to end learning and they are learning scale.
12:01
[SPEAKER_01]: You basically, you show a lot of data that is relevant to the problem you're trying to solve and then you have these algorithms that can look at that data without much oversight and then can turn that into a solution.
12:11
[SPEAKER_01]: People who give you the bullish case on humanoid robots will say this approach, this paradigm has worked in image labeling, it's worked in speech to text, it's worked in text generation in large language models and it's really taken off in these spheres and you can't, you know, for all of the, you know, the complex debates about the utility of large language models, they have accelerated in capability hugely and they have gone from something, you know, I remember playing
12:37
[SPEAKER_01]: I kind of remember that well it's that 2018 that was something like this.
12:42
[SPEAKER_00]: I remember in what was it I think like summer of 22 there was like a moment where like dolly mini images were like going around and like they were they were pretty shit but like you could see that it was getting somewhere right yeah.
12:54
[SPEAKER_01]: Yeah, and so denying, you can't deny that these have solved problems, these paradigms, and the essential belief is that the same thing will hold true for robotics.
13:03
[SPEAKER_01]: The hardware for robotics has always been relatively capable.
13:06
[SPEAKER_01]: There's problems with clusters, problems with speed, problems with stuff like energy supply,
13:10
[SPEAKER_01]: But the hardware, you know, what you can do with an industrial robotics I'm in a factory is amazing.
13:14
[SPEAKER_01]: The precision that you can unleash there, the power is incredible.
13:17
[SPEAKER_01]: So it's mostly being, or a lot about, is the control systems.
13:21
[SPEAKER_01]: It's just incredibly difficult to program a robot to operate in the world with the same dexterity, mobility, as a human does.
13:27
[SPEAKER_01]: these companies are pretty much all of them.
13:29
[SPEAKER_01]: Then to different degrees, you know, they all have slightly different strategies.
13:32
[SPEAKER_01]: They're betting that AI will deliver the control systems that manually programmed control systems could not do.
13:38
[SPEAKER_01]: You're going to feed these systems with a lot of data.
13:41
[SPEAKER_01]: It's going to be collected from video recordings of humans doing things and from recordings of teleoperation.
13:47
[SPEAKER_01]: of people controlling robots like puppets in order to do tasks.
13:51
[SPEAKER_01]: You're going to, you know, cramble that into a transformer or some sort of similar deep learning system and you're going to come out with a general purpose robot on the other end.
13:58
[SPEAKER_01]: That is a huge bet.
13:59
[SPEAKER_01]: That's like, you know, it's not bullshit.
14:01
[SPEAKER_01]: I wouldn't say as an idea, there is a lot of credence to it.
14:04
[SPEAKER_01]: There's a lot of early research that we're seeing, you know, there's companies that are sort of less well known as one called physical intelligence and San Francisco, which does really amazing work here.
14:12
[SPEAKER_01]: Google, DeepMind,
14:17
[SPEAKER_01]: It is a bet that these companies are taking because they believe the human-oid form is going to unlock.
14:23
[SPEAKER_01]: You know, this almost apocalyptic replacement of human labour.
14:26
[SPEAKER_01]: They're promising robots that can do anything humans can do.
14:29
[SPEAKER_01]: And they're promising that within a matter of years.
14:31
[SPEAKER_01]: Now, I want to be clear, that's not going to happen.
14:34
[SPEAKER_01]: It's not going to happen in a matter of years.
14:35
[SPEAKER_01]: Absolutely not.
14:36
[SPEAKER_01]: It may happen over the course of decades that we're going to see slow progress here, but the promises being made by these companies
14:46
[SPEAKER_00]: Yeah, and you know, I think we should talk about Muscle, right?
14:49
[SPEAKER_00]: Because a lot of the recent, like obviously this is clearly benefited from the AI moment, the wave of investment, as you were saying, the belief that these systems can be developed that are going to work for humanoid robots as well.
15:01
[SPEAKER_00]: But it feels like there's no denying that Elon Muscle's well has played a role in
15:07
[SPEAKER_00]: putting energy into this space by trying to do, you know, humanoid robots himself through Tesla, is he like a key kind of mover here?
15:15
[SPEAKER_00]: Yeah, yeah, absolutely.
15:16
[SPEAKER_01]: So for this hard piece, I spoke to a bunch of different executives, and then analysts, you know, you're just happy too.
15:23
[SPEAKER_01]: When reporting this sort of stuff out, and every time I asked them, okay, when was the starting point for this?
15:28
[SPEAKER_01]: To an individual, they all said it was when Musk unveiled the Tesla Bach, which was August 19th, 2021.
15:34
[SPEAKER_01]: Teslabot was the first name for Optimus.
15:36
[SPEAKER_01]: It's the humanoid robot made by Tesla and famously informously we might say Musk did not have a robot on that day.
15:45
[SPEAKER_01]: What he had was an individual in a Spandex robot outfit who got up on stage and danced to dubstep.
15:51
[SPEAKER_01]: Which really dates it I think or I don't know, it dates Elon Musk's cultural sensibilities which have never been cutting edge.
15:58
[SPEAKER_00]: I don't know though.
16:00
[SPEAKER_00]: That could have been a robot.
16:01
[SPEAKER_00]: I think it's a weapon here, right?
16:04
[SPEAKER_01]: But basically, must get some state.
16:05
[SPEAKER_01]: You bring out this guy in a Spandex costume and he says, Look, we're going to build a human level robot.
16:09
[SPEAKER_01]: It's going to have human level hands.
16:11
[SPEAKER_01]: And he makes all of these claims saying, Oh, you're going to be able to make it do any task.
16:14
[SPEAKER_01]: Like go to the store, buy me them out, blah, blah, blah.
16:17
[SPEAKER_01]: These items, and then make a cake and it'll be able to do that.
16:19
[SPEAKER_01]: all of the executive I spoke to.
16:21
[SPEAKER_01]: You know, none of them are going to badmouth muskies.
16:24
[SPEAKER_01]: You know, much of a powerful player.
16:26
[SPEAKER_01]: They're not going to say, oh, he's an idiot.
16:28
[SPEAKER_01]: I mean, some did, but you know, a little in a little gentle terms than that.
16:32
[SPEAKER_01]: But they all said, once he was involved, money rushed in.
16:34
[SPEAKER_01]: Money follows when musk opens his mouth because he is, you know, is he still a world's richest man?
16:40
[SPEAKER_00]: I think so.
16:40
[SPEAKER_00]: Yeah.
16:41
[SPEAKER_00]: I think Ellison is down because Oracle is not doing good and he's up.
16:45
[SPEAKER_00]: Yeah.
16:46
[SPEAKER_00]: We'll see when this gets published, if it's the same way, but at the moment, at least they've been.
16:50
[SPEAKER_01]: Who knows?
16:50
[SPEAKER_01]: He's in the top five, let's say that.
16:52
[SPEAKER_01]: Let's be generous.
16:53
[SPEAKER_01]: So when he gets involved in something, money will follow.
16:57
[SPEAKER_01]: And also for all the criticisms of Moscow, you know, could fill several books, obviously.
17:02
[SPEAKER_01]: He's not a bad judge of when a technology is mature and he saw something like satellite internet and he managed to push that.
17:09
[SPEAKER_01]: He's reusable rockets, debates about the overall utility of that, but he saw that that was a technology
17:15
[SPEAKER_01]: And the same with EVs, he saw that coming early and he was completely failed, you know, rather undermined himself there.
17:22
[SPEAKER_01]: But so he knows when this stuff is going to be coming to the next level.
17:26
[SPEAKER_01]: And so everyone I spoke to you said, Musk has got all the energy in the room.
17:29
[SPEAKER_01]: Now, obviously his own 4A is into this, but the optimist robot have had the usual sort of exaggerated claims.
17:36
[SPEAKER_01]: And many people in a very subtle way I spoke to, you know,
17:40
[SPEAKER_01]: obviously blame Musk for blowing up expectations.
17:43
[SPEAKER_01]: He's he's said he's predicted it's going to be 30 trillion in revenue for Tesla and
17:49
[SPEAKER_01]: from the optimist rate.
17:49
[SPEAKER_01]: It's going to be the world's biggest product.
17:51
[SPEAKER_01]: It's going to eliminate all human label of this absolute pioneer sky nonsense.
17:56
[SPEAKER_01]: And that obviously sets these things up for a full and people don't like that.
17:59
[SPEAKER_01]: So a lot of companies I spoke to, they will say things like, you know, we're glad that Musk has got the energy in here, but we have different goals ourselves.
18:07
[SPEAKER_01]: The big divide in the industry at the moment, basically, is whether you're making humanoid robots to go into the logistics industry, whether they're going to go into the warehouse and move
18:18
[SPEAKER_01]: and that is what a lot of companies are promising at the moment as well.
18:21
[SPEAKER_00]: Before we move on to talking about that, I did want to pick up on what you were saying about Musk, right?
18:25
[SPEAKER_00]: Because it does seem pretty fascinating.
18:28
[SPEAKER_00]: You know, obviously he takes this pivot to talk about going in of robots and how it's going to be this massive new market, which,
18:34
[SPEAKER_00]: I don't know if it's coincidence or what, but it seems to be at the very moment as well that the kind of electric car side of the company seems to be not doing as well facing a lot more serious competition.
18:45
[SPEAKER_00]: And it's been fascinating to see, again, you know, you talk about kind of expectation and hype and what you can get investors to believe.
18:53
[SPEAKER_00]: But every quarter every year you see the kind of core business of this company showing that it is not performing, you know, we see declining sales of Tesla vehicles globally.
19:04
[SPEAKER_00]: But meanwhile like the share price still seems to be in a world of its own.
19:09
[SPEAKER_00]: supposedly enabled by these promises around robots that don't seem to be going anywhere.
19:15
[SPEAKER_00]: Absolutely.
19:15
[SPEAKER_01]: I think Musk, he probably does believe in the technology and believes in some of the promises he's making, but he has always known the utility of showmanship, and he's always been able to find things that gin-up investor confidence, which is the foundation of his wealth obviously Tesla, and its stock price, that is why Musk is such a wealthy man.
19:33
[SPEAKER_01]: And the showmanship, the human version of it, you know, every few months he can say, look, Optimus can do a new trick here.
19:40
[SPEAKER_01]: It can backflip, it can dance, we've got it at the Tron Premier, where it was, you know, it was doing a, you know, karate moves, opposite Jared Lato.
19:47
[SPEAKER_01]: And that is fantastic for Tesla stock price, and he can keep on trickling that out.
19:50
[SPEAKER_01]: You know, I've given a year or two before he puts an Optimus robot on a rocket and sends it up into orbit, or something like he did with the roadster, just because that will generate fantastic headlines.
20:02
[SPEAKER_01]: You know, there's a real plunge in European Tesla sales.
20:04
[SPEAKER_01]: If you're like, right, we're sending optimists to the moon.
20:07
[SPEAKER_01]: And it'll do nothing.
20:08
[SPEAKER_01]: It's like sending a mannequin to the moon.
20:10
[SPEAKER_01]: It'll just fall over when it gets there.
20:11
[SPEAKER_01]: You know, but like the Chinese will get to the moon before the Americans, but at least they'll be a robot up there somewhere.
20:16
[SPEAKER_01]: Yeah, there's that big decaying optimist robot on the moon and the Chinese will go, can we plug this in?
20:21
[SPEAKER_01]: No, I don't think it works.
20:22
[SPEAKER_01]: All right, send it back.
20:25
[SPEAKER_00]: it's so wild, right?
20:25
[SPEAKER_00]: And like we obviously we see that from Musk, you know, you were talking about what these other companies are saying, do you see other ones kind of like really leaning into that hype in order to try to drive investment in things or are many of them trying to be like a bit more cautious with with what they say or is there very much like a divide?
20:42
[SPEAKER_01]: I think trying to be Elon Musk when it comes to outrageous promises on the future.
20:47
[SPEAKER_01]: You know, I think that's a, that's a, that's a suckers game.
20:50
[SPEAKER_01]: No one is going to try and take him on directly because no one else can get away with it.
20:54
[SPEAKER_01]: Basically, yeah, a bit bit you have companies, some of whom are leaning more into that.
20:59
[SPEAKER_01]: I would say the ones that make more bullish claims in the US are figure and one X and that they are.
21:04
[SPEAKER_01]: They're saying we're trying to build a general purpose machine here that is going to replace all forms of human labor.
21:10
[SPEAKER_01]: And the way they see it is, again, dependent on this belief in AI is something that's going to really deliver huge gains over a very short amount of time.
21:21
[SPEAKER_01]: So the pitch you get from the more bullish of these companies is that like,
21:24
[SPEAKER_01]: Okay, you could build something that works in a warehouse, but that's a limited use case.
21:29
[SPEAKER_01]: If we can build something that's going to work in your house, then all of the other jobs are sort of they are built into that in a way.
21:36
[SPEAKER_01]: Once you have that level of dexterity that's capable of navigating and doing useful things in a home, everything else comes under that umbrella.
21:43
[SPEAKER_01]: So they're definitely pushing forward and they're saying we're going to make this big leap.
21:47
[SPEAKER_01]: And then there are companies who would just much more conservative.
21:51
[SPEAKER_01]: I mean, to critics, they're still making unnecessary and unjustified promises, but I think they're much more realistic.
21:58
[SPEAKER_01]: And an example of this would be a company called Agility Robotics, who I also visited, I saw some of their robots in real life talked to their executives, and they are very much focused on logistics.
22:08
[SPEAKER_01]: And that means moving things about in a warehouse.
22:11
[SPEAKER_01]: They have these digit robots, which are not humanoid in their bipedal and they have arms, but they're slightly inhuman in their design.
22:19
[SPEAKER_01]: They have a digit-grade lead, you know, with backward facing knees like you get on ostriches, and they have sort of nubs for hands or very basic claws rather than five finger hands instead.
22:29
[SPEAKER_01]: And I think that's a much more reasonable goal because it's a logistics in the US, particularly, is a really understaffed industry because for one thing, it's very, very difficult work.
22:39
[SPEAKER_01]: It's backbreaking work.
22:40
[SPEAKER_01]: It's not the best paid in the world.
22:42
[SPEAKER_01]: So this is a industry which is really actively looking for.
22:46
[SPEAKER_01]: to use a euphemism solutions, right?
22:49
[SPEAKER_01]: And they are very keen to test out these robots.
22:51
[SPEAKER_01]: But even with these companies that are mostly looking logistics, and this is something that I'll probably stress a few times, the numbers here are tiny.
22:59
[SPEAKER_01]: The numbers are so small about the actual humanoid robots being deployed, just for some context here.
23:05
[SPEAKER_01]: There are about
23:06
[SPEAKER_01]: 4.2 million industrial robots, currently an operation across the globe.
23:10
[SPEAKER_01]: When I mean industrial robots, I mean big robot arms that are doing heavy lifting, that are doing welding, paint jobs, all this sort of thing.
23:17
[SPEAKER_01]: All of these companies in the US, when they say we've got a deployment and we've got it with BMW, we've got it with Amazon, you're looking at maybe two or three robots in the factory.
23:26
[SPEAKER_01]: And they're not, you know, they're not doing full-time work, they're not integrated into the workforce, like anything else, these are pilots.
23:33
[SPEAKER_01]: The last year, the company that shipped the most humanoid robots was unitary, Chinese company.
23:38
[SPEAKER_01]: And they've done fantastic in terms of volume, but they only shipped five and a half thousand robots over the entire year.
23:44
[SPEAKER_01]: But this is 4.2 million deployed across the world.
23:47
[SPEAKER_01]: So whenever we talk about this, we've just got to remember that we're talking about really tiny figures.
23:51
[SPEAKER_01]: And even if the utility is there, the adoption is a long way to go.
23:57
[SPEAKER_00]: If you've ever thought, wow, that scammymail feels weirdly personalized.
24:00
[SPEAKER_00]: That's not intuition.
24:01
[SPEAKER_00]: That's the data economy working exactly as intended.
24:04
[SPEAKER_00]: That's why I want to talk about incognito.
24:06
[SPEAKER_00]: When your home address, phone number, email, employment history, and other personal details are sitting in searchable databases, it increases your exposure to fishing attacks, identity theft, and financial fraud.
24:16
[SPEAKER_00]: This isn't just about annoying spam, it's about real risk.
24:19
[SPEAKER_00]: Incognito works to remove you broadly from the internet.
24:22
[SPEAKER_00]: They identify where your data appears and send legally binding or removal requests under laws like GDPR and CCPA.
24:28
[SPEAKER_00]: Follow up until Deletion is confirmed and continuously monitor.
24:31
[SPEAKER_00]: So it doesn't quietly reappear later.
24:33
[SPEAKER_00]: And what I really like is their customer removal feature.
24:36
[SPEAKER_00]: If your information shows up on a specific site outside their automated system, their team will manually pursue that removal too.
24:43
[SPEAKER_00]: That's included with the unlimited and family unlimited plans, and honestly, if you want the most comprehensive protection, especially if you're thinking about your household, that's the route I'd recommend.
24:52
[SPEAKER_00]: This is about shrinking your digital footprint and lowering your exposure to scams and identity theft and not pretending the problem doesn't exist.
24:59
[SPEAKER_00]: Right now, you can get 60% off an annual incognit plan.
25:02
[SPEAKER_00]: Go to incognit.com slash save us and use code save us, SAVEUS.
25:07
[SPEAKER_00]: You can't dismantle surveillance capitalism overnight, but you can make yourself a less convenient data point.
25:12
[SPEAKER_00]: Again, that's incognito.com slash save us, code save us for 60% off today.
25:19
[SPEAKER_00]: It seems to me when you're talking about robots in a warehouse, like the thing that that immediately comes to mind is 20 years ago or whatever, we have Amazon moving into warehousing in the United States in particular and bringing down the wages by trying to bust the unions in logistics.
25:35
[SPEAKER_00]: And now it feels like there were these stories a few years ago that they were running out of labor, you know, they were having a hard time trying to find enough people because, you know, they're kind of labor churn through an Amazon warehouses tends to be quite high.
25:47
[SPEAKER_00]: And so it doesn't seem surprising to me that there would be a greater demand for robots to do more work.
25:54
[SPEAKER_00]: And of course, we know that Amazon has had robots in there for quite some time, but, you know, what the future of that might actually look like right.
26:01
[SPEAKER_01]: Yeah, I mean, it lambs in such an interesting case because no to be, it doesn't have a human on a robot, right?
26:07
[SPEAKER_01]: It's definitely researching in areas like this, but the robots that have been responsible for a large degree of automation in Amazon warehouses or what's known as AMRs, autonomous mobile robots, and these are small things that basically look like coffee tables and wheels,
26:21
[SPEAKER_01]: and these are the ones that sort of scoot around and they lift up under a stack of shelves and they move that to another destination.
26:27
[SPEAKER_01]: And then it's the human that does the difficult dexterous tasks of picking out products, putting them in a package, wrapping it up, sending it enough.
26:34
[SPEAKER_01]: So Amazon, which you know, more capital than anyone and more incentive than anyone to automate this business as much as possible, they don't have you much.
26:41
[SPEAKER_01]: I'm that of course they're looking at it, but that's something that is just it's not there yet.
26:47
[SPEAKER_01]: And I think that, you know, the point you make about the automation of these of their warehouses, obviously this is a much bigger conversation.
26:53
[SPEAKER_01]: But this is definitely a motivation, especially for people like Musk.
26:57
[SPEAKER_01]: You know Elon Musk hates unions.
26:59
[SPEAKER_01]: He has battles with them constantly.
27:01
[SPEAKER_01]: And I think a big part of the appeal of a humanoid
27:04
[SPEAKER_01]: It's not just the shock factor, it's not just the utility, but it's the idea of having a completely client frictionless workforce.
27:11
[SPEAKER_01]: And that is another reason why there's such buy-in from elite and financial elites because they're like great.
27:16
[SPEAKER_01]: We can get rid of the pesky humans who keep on going up our perfect economies and we can replace them with robots instead.
27:22
[SPEAKER_01]: But that's I think that's a bigger topic in a way.
27:25
[SPEAKER_00]: Of course, but even as you say that though, like it brings to mind, I thought that I was having, as I was reading your piece, which was very much that like, to certain people developing these robots, it does feel like there is a desire, and we've heard this criticism for a long time when it comes to robots, to basically create the acceptable slave, to have the robot that can be completely controlled, or the humanoid form that can be completely controlled, and now you don't need to worry
27:52
[SPEAKER_00]: what is actually happening to the human because you can do whatever you want to to the robot.
27:56
[SPEAKER_00]: But again, like if the robot is not really coming, but yeah, I don't know.
28:00
[SPEAKER_00]: It's like the desire at least among some of them seems a bit questionable or worrisome, I guess.
28:06
[SPEAKER_01]: Yeah, it was actually something I did write about for Harper's and you know, got left on the edit editing floor is a lot of a lot of things do, because it's it's difficult to speculate without coming across has, I don't know, you know, it are these people building humanoid robots because they secretly long to dominate and control humans a lot of people would say yes aptly
28:27
[SPEAKER_01]: Absolutely, that's what they're doing.
28:28
[SPEAKER_01]: You ask someone who's doing this and they go, God, no, further thing from my mind, I'm all about, you know, creating a utopia.
28:34
[SPEAKER_01]: Is that part of people's motivation?
28:36
[SPEAKER_01]: I mean, probably on some subconscious level, I've would imagine that is a desire to be like, I want to have a human form that I can control.
28:44
[SPEAKER_01]: I think there is a really interesting conversation here about stuff like giving rights to robots how they should be treated.
28:50
[SPEAKER_01]: You know, I'm one of these people who is very anti robot rights as it were because I believe it's essentially a degradation of human rights.
28:58
[SPEAKER_01]: It doesn't elevate robots to the same status as humans.
29:02
[SPEAKER_01]: It degrade humans to the same status as robots.
29:05
[SPEAKER_01]: And I think we should always when we're looking at these machines, treat them as what they are, which is avatars of capital.
29:11
[SPEAKER_01]: You're not being nice to a conscious being if you help a robot across the street, you are helping capital in some way.
29:16
[SPEAKER_01]: You know, and I think we're very susceptible to being manipulated for the same reason that I've talked about over exaggerations of capability.
29:23
[SPEAKER_01]: We see a human form and we start treating it like a human.
29:26
[SPEAKER_01]: I think that's very dangerous and it's something that's going to be a real conversation.
29:31
[SPEAKER_01]: We're really going to have to consider seriously over the coming decades.
29:35
[SPEAKER_00]: I think it both says something actually really good about, you know, what is kind of like intrinsic to humanity, right?
29:43
[SPEAKER_00]: you know we see this thing we want to help it because that's kind of like built into our nature even though we're often told that we're very competitive beings and all this kind of stuff but then at the same time you can see how companies can very much take advantage of that you know in transit part of our nature to benefit the sense
29:59
[SPEAKER_00]: even on that point though, and there are some things that I wanted to go back to that that you were saying, but since we're on the subject, talking about the labor piece, talking about the desire for control, it feels like there's another step here to take where, obviously there's the desire to control the robot, but there are also all of these discussions about what is actually
30:22
[SPEAKER_00]: And, you know, we see this with the AI conversations in the chat box and stuff as well, or whether it's the autonomous cars and all this kind of stuff, but does the kind of form or the idea that the computer is doing it then allow for greater control and subjugation of labor that's hidden away behind it all, and I know that there are discussions in the kind of humanoid robot spaces to how these things are being trained, whether they're being remotely controlled.
30:47
[SPEAKER_00]: Can you talk a bit about that aspect of this and what we're seeing there?
30:51
[SPEAKER_01]: This is something that's particularly come about with humanoid robots that are being deployed in the home.
30:56
[SPEAKER_01]: The leader in that space is a company called Onex or in the US at least, and they have a robot called the Neo that they're saying, we're going to put it in the home and then we're going to learn how to do all these tasks as part of that.
31:07
[SPEAKER_01]: They went up for pre-order last October, I think, for $20,000, and they've done some recent demonstrations.
31:13
[SPEAKER_01]: There was a great video with Joanna Stern looking at it doing tasks in the home and then taking a dishwasher.
31:18
[SPEAKER_01]: They're promising that this stuff is going to be autonomous, that the AI, the CHAPGPT approach is going to create a training system.
31:25
[SPEAKER_01]: It's simply not there yet, and instead they have what's called teleoperation, which is, as I mentioned earlier, you're sort of controlling the robot-like puppet.
31:33
[SPEAKER_01]: You have a pair of VR glasses on that gives you views out of its cameras, and you have gloves, or you have some sort of controller that mimics the hand motions.
31:45
[SPEAKER_01]: It's a little bit of a trick, really, isn't it?
31:47
[SPEAKER_01]: The hope is that you do that enough and then you collect the data and then you can program an AI system to do that.
31:52
[SPEAKER_01]: But at the moment, anyone who's buying or who tries to get one of these robots in their house, or what's going to have a person in their house, it's going to be a person who's working for one ex, working remotely, and is getting to snoop around their house in some way, and actually have a physical present in it.
32:05
[SPEAKER_01]: I don't think we're at the point where,
32:08
[SPEAKER_01]: we are seeing a sort of mass hidden labor force in the same way that the AI world, you know, it's be, you know, widely reported that you have these data labeling centers in countries with poor incomes than the US and Canada and places like this.
32:24
[SPEAKER_01]: And, you know, they are the human workforce behind AI.
32:27
[SPEAKER_01]: That's not something we're currently seeing with robots simply because the industry is not at the same scale, but it is something that could happen in the future.
32:34
[SPEAKER_01]: I think one possible future we're going to see for humanoid robots is that the AI control systems are not going to be safe and reliable enough in order to have these robots do things autonomously, but the hardware will be there and we're going to see bigger teleoperation farms, let's say.
32:51
[SPEAKER_01]: So something like a core centre where you offshore that labour to another country and you get people overseeing perhaps one perhaps multiple robots at a time and they'd be ready to step in.
33:02
[SPEAKER_01]: I think that creates it's really fascinating in very grim labor problem where rich homes are going to say, well, I've outsourced my labor to this robot, and this way I don't even need to see the person who's doing this work for me.
33:17
[SPEAKER_01]: And it really reminds me of the story we get about the invention of the done waiter, which was something that seems like in a way a technological marvel, but a big part of the motivation for making the done waiter was that people didn't want to see
33:32
[SPEAKER_01]: that was the era that it came about and it done way to was a way of hiding that work force.
33:37
[SPEAKER_01]: And I think that's something we could end up seeing with humanoid robots as well.
33:40
[SPEAKER_00]: Yeah, definitely.
33:41
[SPEAKER_00]: It brings to mind in the kind of worst case scenario.
33:44
[SPEAKER_00]: I don't know if you've seen this Mexican science fiction film a sleep dealer from 2008 to 2009.
33:49
[SPEAKER_00]: Okay, but it's basically about like,
33:51
[SPEAKER_00]: you know, the border is closed between the U.S. and Mexico, but the U.S. still relies on all this kind of Mexican labor, but they just go into these big centers and kind of teleoperate these robots across the border.
34:03
[SPEAKER_00]: And it's like, it feels like we're a heading closer closer to that space.
34:07
[SPEAKER_00]: And I spoke to the director, and he even told me he was talking to like tech CEOs in the past, and they were like, yeah, we're really inspired by your movie instead.
34:18
[SPEAKER_01]: Oh, yeah, we built the next installment, the tourment nexus, exactly.
34:22
[SPEAKER_01]: Yeah, exactly.
34:23
[SPEAKER_01]: Yeah.
34:24
[SPEAKER_01]: I had heard of that film.
34:24
[SPEAKER_01]: That sounds fantastic.
34:26
[SPEAKER_01]: Yeah.
34:26
[SPEAKER_01]: I mean, I think that's like, and that ties into so many strands of international politics by seeing all these desires are closed borders to become fortress nations.
34:35
[SPEAKER_01]: turn policies on immigration and something I'm from the UK.
34:38
[SPEAKER_01]: It's such a huge part of the political picture here.
34:41
[SPEAKER_01]: And I think that the people who are building these robots, they're not explicit about that.
34:45
[SPEAKER_01]: No one is saying, hey look, build our robots and you won't need to have immigrants living in your country doing this manual labor.
34:51
[SPEAKER_01]: But I'm sure that that crops up in conversations.
34:53
[SPEAKER_01]: You know, it's Davos is happening right now.
34:55
[SPEAKER_01]: I'm sure in some unrecorded background people will be saying, well, this solves the political tensions of immigration.
35:01
[SPEAKER_01]: So yeah, I think that's definitely that's going to be part of the picture as well unless you're Alex Carp of Paladir and then you'll just say it out in the open Yeah, I was going to say actually one person who could match Elon Musk in terms of emotional and overwrought and substantiated outbursts would be Alex Carp
35:17
[SPEAKER_00]: Yeah, that's really agree with that.
35:19
[SPEAKER_00]: But okay, I do want to kind of backtrack a little bit, right?
35:22
[SPEAKER_00]: Because a few times you brought up how much this is related to generative AI and what we have seen there.
35:29
[SPEAKER_00]: And so I'm sure that this is an obvious question, but I just kind of want you to state it for the audience for people listening and might not be as tuned in.
35:36
[SPEAKER_00]: We're not really talking about large language models going into these.
35:39
[SPEAKER_00]: These robots are we or is this kind of like using the tools that help to create large language models to now create different forms of AI that are helping to create these robots.
35:49
[SPEAKER_01]: So it's not that you're putting large language models in these robots but you're using some of the same paradigms and that's mostly about things like data scaling and getting a lot of information and using end-to-end learning to get a training model without having to be explicit about what you're training to do.
36:05
[SPEAKER_01]: I mean some companies are adding LLMs to robots but they're just as an interface.
36:10
[SPEAKER_01]: You know, so you can put a little agent inside a robot and that way you can talk to it.
36:16
[SPEAKER_01]: You can have a speech, text a speech thing, and you can give it instructions that it then tries to interpret.
36:21
[SPEAKER_01]: But that sort of a layer on top, like a UI layer rather than something that's fundamentally apparent.
36:26
[SPEAKER_00]: I appreciate O-learning it just so just so people are aware.
36:29
[SPEAKER_00]: In terms of, again, you've mentioned before that we have a bunch of robots around us, right?
36:34
[SPEAKER_00]: Already, they just don't take the humanoid form that these companies are trying to pioneer.
36:40
[SPEAKER_00]: So what is the argument for trying to do robots as humans versus doing robots that are tailored to specific tasks and don't try to replicate what we look like?
36:51
[SPEAKER_01]: Yeah, I mean, the thing that you hear over and over again is that the world is built for humans.
36:56
[SPEAKER_01]: So let's build a robot that fits the environment that already exists rather than having to adapt the environment to the robot itself.
37:03
[SPEAKER_01]: It's a very good argument.
37:05
[SPEAKER_01]: You know, there's the world is full of steps, stairs, handles, levers, scissors that you need to hold knives that you need to hand to hold.
37:11
[SPEAKER_01]: Well, sort of need to hand to hold.
37:12
[SPEAKER_01]: and that's why building a human shaped robot is basically a good idea.
37:16
[SPEAKER_01]: But again, that is sort of predicated on the belief that we can reach something approaching human level, dexterity and mobility, which I just don't think is going to happen in the short term.
37:27
[SPEAKER_01]: And then there's two important things to think of when we're talking about why shaped like a human.
37:32
[SPEAKER_01]: one is really understanding human dexterity and why it is so incredible.
37:36
[SPEAKER_01]: I was doing, you know, a little bit of reading before this interview, like brushing up on stuff that I thought about, and I came across this amazing video, linked to by Robinie Brooks, this guy I mentioned earlier, which was of some tests done by a Swedish researcher called Roland Johansson.
37:51
[SPEAKER_01]: in which he got volunteers to light a match and he did it once normally and you are okay great I can let him act to spine and for the second time round he enneathed the ties their fingers so he took all of the sensation out of their fingertips and then said right now light a match again and this video is amazing because you see these people
38:08
[SPEAKER_01]: reaching for the matchbox and sort of like grabbing a handful of matches and going, oh no, how do I do it and dropping them and picking them up again?
38:15
[SPEAKER_01]: And then you realize the amount of sensory data that is built into humans and our understanding of the world is immense.
38:23
[SPEAKER_01]: And that's something that we're just not capturing in these robot training things at the moment.
38:27
[SPEAKER_01]: So I've talked about
38:28
[SPEAKER_01]: using teleoperation to create training data in order to train the robots.
38:33
[SPEAKER_01]: That's the, for example, is doing this with vision only, the same approach it's taken with its cars.
38:37
[SPEAKER_01]: It is not incorporating force feedback, pressure stuff.
38:40
[SPEAKER_01]: I'm sure that, you know, they'll change that, that policy will change and they're being a little bit opaque about it, but that's the latest report.
38:46
[SPEAKER_01]: So you're missing out on all this data and you are like a person with a neuthotized finger tips, funbbling for a matchbox.
38:51
[SPEAKER_01]: That level of dexterity is just so hard to do.
38:54
[SPEAKER_01]: Current robots, they usually have about a precision level of one to three centimeters.
38:59
[SPEAKER_01]: That's the sort of margin of error when they're picking things up.
39:02
[SPEAKER_01]: If you think about, if you were to cook a dish in your kitchen and you had a three-centimeter margin of error in every grass you took, every cut you were doing on a carrot, you would mess things up pretty quickly.
39:15
[SPEAKER_01]: Get down to one centimeter.
39:16
[SPEAKER_01]: You're doing pretty good.
39:17
[SPEAKER_01]: You're probably like a a child.
39:19
[SPEAKER_01]: You're going to fumble some things, you're going to drop stuff, but you'll maybe get there.
39:22
[SPEAKER_01]: but you're still going to be unable to do even trickier tasks as a great sort of discussion about what is the real robot Olympics?
39:31
[SPEAKER_01]: Because we see these robot Olympics when they're doing like running and they're kickboxing and they're playing football and the real robot Olympics, some of these new roboticists say is okay can you put on a latex glove?
39:41
[SPEAKER_01]: Can you peel an orange?
39:43
[SPEAKER_01]: Can you shave?
39:44
[SPEAKER_01]: You know all of these things and these are things that humans take for granted and we're just
39:50
[SPEAKER_01]: So that's one point in the human form, they're claiming that you can therefore fit the robot to the environment and you can unlock on all these tasks and the criticism that is actually the nature of that unlock is so much more difficult than we think.
40:02
[SPEAKER_01]: A corollary of that is, okay, humanoid robots may be not going to take off in the way that these companies are promising.
40:08
[SPEAKER_01]: But there's no reason there shouldn't be other morphologies.
40:10
[SPEAKER_01]: And we see lots of companies experimenting with this.
40:13
[SPEAKER_01]: So for example, there's a company in the US called Sunday Robotics.
40:16
[SPEAKER_01]: And they do a two-armed robot with a sort of head sensor, but it's on wheels instead.
40:20
[SPEAKER_01]: And so it just gets rid of the problems of legs.
40:22
[SPEAKER_01]: It gets rid of the problem of robot toppling over or the power shutting off and it suddenly, you know, the robot falls down and it crashes your dog or whatever it might be.
40:30
[SPEAKER_01]: That's a really good way to do it.
40:32
[SPEAKER_01]: I've seen some amazing videos of automated toilet cleaning.
40:35
[SPEAKER_01]: Sorry, I just heard myself say that sentence out loud.
40:38
[SPEAKER_00]: Ha ha ha!
40:39
[SPEAKER_00]: It's quite a dull one, isn't it?
40:42
[SPEAKER_00]: You're watching a lot of videos about Japanese toilet seats or something like that.
40:47
[SPEAKER_01]: Yeah, I'd want to, well, you know, you got it, it's free search.
40:49
[SPEAKER_01]: Yeah, but like you get these automated toilet cleaners.
40:51
[SPEAKER_01]: And they're like little caddies on wheels.
40:53
[SPEAKER_01]: And they have one industrial arm.
40:55
[SPEAKER_01]: It's flexible in ways of human arm isn't.
40:57
[SPEAKER_01]: And you have an array of sensors.
40:58
[SPEAKER_01]: One, it does.
40:59
[SPEAKER_01]: It scans all of the surfaces.
41:00
[SPEAKER_01]: And then there are just scrubs everything.
41:02
[SPEAKER_01]: And it moves around the little wheels.
41:04
[SPEAKER_01]: That's really useful.
41:05
[SPEAKER_01]: That I can see appearing in public bathrooms in the next 10 years.
41:09
[SPEAKER_01]: I think a lot of robotics we have in the world today is sort of invisible because it's not human-shaped.
41:15
[SPEAKER_01]: You do something like, I don't know, you transit through a major airport.
41:19
[SPEAKER_01]: When I was doing this reporting trip, I was going through London, he throw and there were robot cleaners doing the floors.
41:24
[SPEAKER_01]: They're just these little trolley things with, you know, mobs, heads and brushes underneath, but that's a robot that is already become commonplace and I think that would look futuristic
41:35
[SPEAKER_00]: It's a good point and I was wondering like thinking about what is futuristic and and what we expect.
41:41
[SPEAKER_00]: Like I wonder how much of this desire to build the humanoid robot and to roll them out is actually being realistic about what is possible in the moment and how much of it is like again, thinking back to conversations that we've been having about.
41:55
[SPEAKER_00]: catchy BT in generative AI, like how much is it like kind of a faith-based argument that like this is, how the future should work.
42:03
[SPEAKER_00]: This is what technology should be, and so of course we need to like put all these resources into realizing it.
42:09
[SPEAKER_01]: I think that's always part of it.
42:10
[SPEAKER_01]: And again, it's something that like
42:12
[SPEAKER_01]: people will rationalize and they will come up with post-hot rationalizations when you ask them about this.
42:18
[SPEAKER_01]: But it's clear, you know, must being a good example of this.
42:20
[SPEAKER_01]: Well, you know, the frequency with which he references science fiction.
42:24
[SPEAKER_01]: And I think he even said recently, there was some recent comment he made, where he was promising, you can all have a CP30 of your own.
42:33
[SPEAKER_01]: Of course, of course.
42:35
[SPEAKER_01]: I'd rather have an RTD to myself.
42:37
[SPEAKER_01]: I think the beats are very endearing.
42:39
[SPEAKER_00]: See, three people is like a bit annoying.
42:41
[UNKNOWN]: Yes.
42:41
[SPEAKER_01]: Yeah.
42:42
[SPEAKER_01]: Bumbling.
42:43
[SPEAKER_01]: I don't know if he could do a good job in the kitchen.
42:45
[SPEAKER_01]: I think he'd be knocking things over.
42:47
[SPEAKER_01]: But this is another part of the cultural outreach.
42:51
[SPEAKER_01]: And often when taxi-eos, they refer to this stuff because it provides a common ground between what people think is possible and what is actually possible.
42:59
[SPEAKER_01]: you know, and bridges that gap and it helps the imagination come alive.
43:03
[SPEAKER_01]: So I think that is it's a big part of it and I don't think it's necessary, realistic vision of the future for that reason.
43:10
[SPEAKER_00]: I feel you and on that point though, we have been talking a lot about the United States and I guess to a lesser degree Europe right and what we're seeing there and I feel like a lot of these discussions around where humanoid robots are going has centered a lot on China actually and what has been happening over there.
43:27
[SPEAKER_00]: So what is the state of, I guess, the human robotics industry in China at the moment?
43:33
[SPEAKER_00]: And how is it distinct from what we're seeing in St. North American Europe?
43:37
[SPEAKER_01]: China is absolutely on pace to outstrip the US and Europe is not really in the conversation to be honest.
43:43
[SPEAKER_01]: US is the second place here, definitely.
43:46
[SPEAKER_01]: The China is on outpace to outstrip America in this by the end of this century no doubt at all.
43:52
[SPEAKER_01]: there's a number of different factors involved here.
43:54
[SPEAKER_01]: Why is China going to be better at this?
43:56
[SPEAKER_01]: Partly it's state backing.
43:57
[SPEAKER_01]: The state is, you know, has a lot of economic leverage and it's decided that humanoid robotics are a key part of its future industrial strategy.
44:06
[SPEAKER_01]: So they're throwing a lot of money behind it, they're putting in a lot of orders and they can soak up a lot of redundancy.
44:10
[SPEAKER_01]: In terms of, you know, units being produced aren't necessarily useful but that help a company develop their technology.
44:16
[SPEAKER_01]: The other thing is that these Chinese companies are where all these components are coming from anyway.
44:21
[SPEAKER_01]: You look at any robot that's built in the US, sorry, and the vast majority of the components will be Chinese.
44:27
[SPEAKER_01]: These Chinese companies, they're all next door to each other.
44:29
[SPEAKER_01]: This really helps with prototypeing.
44:31
[SPEAKER_01]: If you have a robot that you're trying to develop, and you're like, well, this part doesn't quite work.
44:35
[SPEAKER_01]: I need to do this rather than waiting two, three weeks for those units to be shipped halfway across the world.
44:40
[SPEAKER_01]: And back, you can just go down the road and you can get that fixed in a couple of days.
44:44
[SPEAKER_01]: So all of these things, this production of scale, the state backing, the supply lines, the ecosystem, all of these are in China's favor.
44:51
[SPEAKER_01]: Now, what is the difference between Chinese and US robotics at the moment?
44:55
[SPEAKER_01]: The people I spoke to would give the edge to the US on the quality of the software and the reliability of the units.
45:02
[SPEAKER_01]: They say the US is much better at building user interfaces that make it easy to actually get these robots working in facilities.
45:09
[SPEAKER_01]: And when I say that, I'm not talking about humanised yet, obviously, as I've sort of pointed out already, those are only being tested in very small numbers, but you look at something like Boston Dynamics quadrupedal robot spot.
45:20
[SPEAKER_01]: That is something that's being used pretty widely now in tasks like industrial inspection, and Boston Dynamics has a fantastically capable software array that lets you just basically use spot like you would control a character in a video game, it's very intuitive, it's very easy.
45:36
[SPEAKER_01]: Most people I spoke to said the Chinese robots, the Chinese robotic factories and suppliers, they just couldn't do that yet.
45:41
[SPEAKER_01]: There's also a problem with quality supply.
45:43
[SPEAKER_01]: I spoke to a U.S. robot reseller who sounds a load of Chinese units to the U.S. market from companies like Unitary and Uptech.
45:50
[SPEAKER_01]: And he said about a quarter over the Chinese robots that he bought in had something wrong with them.
45:56
[SPEAKER_01]: Sometimes it was something simple, like a few wires that weren't missing.
46:00
[SPEAKER_01]: And sometimes it was actually that an entire sensor array was broken and needed to be switched
46:04
[SPEAKER_01]: there's a quality difference there as well.
46:07
[SPEAKER_01]: Now, which of those two approaches is going to rip when out in the end?
46:10
[SPEAKER_01]: You know, you can classify the US as an iPhone approach, just going for high end, top quality, user experience, and you can classify China as the Android approach.
46:19
[SPEAKER_01]: It's going for mass market, it's going for cheap, it's going for volume, and it's going for a lot of different configurations.
46:24
[SPEAKER_01]: I think in the end China is going to control most of the market.
46:29
[SPEAKER_01]: We've seen this already with other sort of big new technologies of the century electric vehicles, solar panels, drones, quadcoctors, China has managed to dominate the stuff.
46:40
[SPEAKER_01]: the sort of comparison tail is often DJI for a while in the 2010s the US had this domestic quadcopter industry and they were just outcomputed by DJI which was offering better or same or better specs but for lower costs because they were able to produce at higher volumes.
46:57
[SPEAKER_01]: Once they've taken the market share and their competitors have shattered they can then tinker around with that and start making a little bit more money on their products if they want to.
47:10
[SPEAKER_01]: Now, the interesting thing about that, I think, is that a lot of analysts I spoke to said that the threat, as it were, as they saw it, of Chinese automation, was really what was inspiring, a lot of interest in humanoid robotics from governments like the U.S.
47:26
[SPEAKER_01]: because one analyst I spoke to you compared it to the Cold War.
47:29
[SPEAKER_01]: They said that this was the new, it was the industrial Cold War.
47:32
[SPEAKER_01]: It's going to be China versus the US and the first country that can print its own workforce as it were.
47:38
[SPEAKER_01]: They are going to capture the entirety of the global economy in the 20th century in beyond.
47:43
[SPEAKER_01]: So I think it's become almost in people's minds and existential question.
47:47
[SPEAKER_01]: Now, I don't think it's as serious as that, because I think the quality of this technology in the adoption is going to take longer than these people for sea, but that's another reason why the human form has this particularly compelling aspect to it, because it makes people think of, you know, legions of work is stamping out of Chinese factories and taking jobs from everyone.
48:05
[SPEAKER_01]: So that's another thing to factor into these discussions that you see.
48:09
[SPEAKER_00]: It's a very useful framing, especially if you're trying to get government support to make sure that your height-based industry is going to still have some runway, right?
48:19
[SPEAKER_00]: In the same way that we see China is going to be the U.S. on AI, or it's going to get AGI first as being like an important narrative to justify government defense of an investment in the AI industry.
48:32
[SPEAKER_01]: And one thing I would pick up from that, something we've not mentioned so yet, yes, is industrial safety standards.
48:38
[SPEAKER_01]: Again, these words come out my mouth and I think, God, that's boring James, but no, it's fascinating.
48:42
[SPEAKER_01]: You know, because a big challenging adoption is going to be, how do we make sure these robots are safe for humans to work around?
48:48
[SPEAKER_01]: At the moment, they're just not.
48:49
[SPEAKER_01]: They fall over, they could do that without warning, they could really hurt someone, they could crush hands, all of this sort of stuff.
48:55
[SPEAKER_01]: And companies and individuals, some of them are arguing, well, let's lower Labour safety standards.
49:01
[SPEAKER_01]: Let's make it less safe.
49:03
[SPEAKER_01]: And now then that is something that obviously you see why that has an appeal to someone like Elon Musk.
49:07
[SPEAKER_01]: Because he's always going on about OSHA safety standards and how these are terrible things for Tesla.
49:12
[SPEAKER_01]: And if you can say, well, hey, if we don't lower safety standards, the Chinese are going to beat us to this.
49:16
[SPEAKER_01]: That's going to be an incredible motivating factor, a justification.
49:20
[SPEAKER_01]: So I think that's another thing we need to think about when we're talking about these robots in the workplace.
49:24
[SPEAKER_00]: I think a very good point and not one that I even considered.
49:27
[SPEAKER_00]: I was wondering on the broader point about China.
49:30
[SPEAKER_00]: I mean, we know that China has been moving forward a lot with kind of robotics technologies generally.
49:36
[SPEAKER_00]: You were talking about earlier, you know, the factory automation and the types of things that we're starting to see.
49:41
[SPEAKER_00]: in China around that and the advances that it's making in that field.
49:45
[SPEAKER_00]: So I guess I'm wondering, we're talking about a humanoid robotics industry in China.
49:50
[SPEAKER_00]: I wonder, first-hand, does that benefit from the broader robotics industry that's going on there?
49:55
[SPEAKER_00]: But secondly, I imagine that this attempt to do humanoid robotics must just be like really a footnote on this broader conversation that has much more attention on kind of factory
50:09
[SPEAKER_01]: Yeah, I mean, on your first point, absolutely there's huge amounts of overlap, the components that you're building in order to solve, let's say, traditional industrial problems, you know, you can take those, you can take those controls systems, you're making, you can take those manipulators and you can put them in a different form factor, so they overlap between this technology is a huge and it's very, very useful to China and.
50:29
[SPEAKER_01]: I agree with the other point, you know, I gave this, that it's a footnote to these wider industrial automation strategies.
50:35
[SPEAKER_01]: I gave this example earlier of unitary putting out 5,500 robot.
50:39
[SPEAKER_01]: That's obviously such a drop in the ocean compared to China's entire strategy.
50:44
[SPEAKER_01]: So I think, you know,
50:46
[SPEAKER_01]: It's a really useful thing to help keep this all in proportion, that even China, which is, you know, has rapidly industrialised rapidly automated, even they're still making pretty small numbers of these robots.
50:57
[SPEAKER_01]: And I would say there's a difficulty here again with the the opacity of the situation, understanding what these companies are really doing.
51:04
[SPEAKER_01]: So,
51:05
[SPEAKER_01]: It's difficult for me and it's difficult for reporters in the U.S. and Western nations in general to get a real clear look inside of Chinese factories and Chinese systems.
51:14
[SPEAKER_01]: I had some interviews with Chinese companies and getting hold of someone who was not their media cons person was basically possible.
51:23
[SPEAKER_01]: You're not getting to speak to people who are deploying this technology.
51:27
[SPEAKER_01]: I think I should take the fault for that for some degree.
51:29
[SPEAKER_01]: Obviously, I am not a journalist covering China,
51:34
[SPEAKER_01]: But whenever we say, oh, let's look at with suspicion on what countries in North America and Europe are doing, we shouldn't forget to apply that suspicion to China as well.
51:45
[SPEAKER_01]: You know, skepticism does not just apply to the US.
51:48
[SPEAKER_01]: It applies to what China is saying as well.
51:50
[SPEAKER_01]: And they have just as much emotion motivation to exaggerate the capabilities of their robots as industrial leaders in the US too.
51:56
[SPEAKER_00]: It brings to mind something that I was thinking again, like as I was reading your piece and trying to get an idea of where this is all going, and it brought to mind the 2010s, right?
52:05
[SPEAKER_00]: When we had this other moment where AI and robotics were supposedly on this cusp of taking off and all of these jobs were supposedly going to be automated and removed and there was even talk of.
52:19
[SPEAKER_00]: rolling out robots in fast food places and in elder care and, you know, in all these kind of places where I feel like Once again, we're hearing discussions of robots being created and deployed into and I wonder if we're just kind of seeing a repeat of waves of of hype that we have seen in the past or What you think are potential real threats that that we do see in this moment even if things might be exaggerated
52:48
[SPEAKER_01]: I worry that the answer I'm going to give is just sort of too moderate, which is that we are seeing a repeat of hype, but there has been a step change in the technology as well.
52:57
[SPEAKER_01]: I think back to the 2010s when, for example, the DARPA Robotics Challenge, I don't know if you ever remember watching footage from that, you know, this was...
53:05
[SPEAKER_01]: sponsored by DARPA and this was asking robots to do things like walk through doors, you know, real difficult stuff at the time.
53:12
[SPEAKER_01]: And I remember writing about that and seeing robots just falling over, grasping a handle and just immediately toppling backwards, that sort of stuff is really easy for robots to do now.
53:22
[SPEAKER_01]: You know, there has been huge amounts of progress, whether that translates into a machine that is capable of replacing human labor.
53:32
[SPEAKER_01]: No.
53:32
[SPEAKER_01]: not right now.
53:33
[SPEAKER_01]: 10, 20 years time.
53:34
[SPEAKER_01]: Yeah, let's have another look and it's see if it's doing anything different.
53:38
[SPEAKER_01]: But we can't always dismiss the fact that the tech industry operates this incredibly efficient and manipulative hype machine.
53:46
[SPEAKER_01]: We can't use that to dismiss the fact that technological changes are taking place.
53:50
[SPEAKER_01]: The robots I saw in person, they were genuinely impressive to me and I saw them doing I don't know if you call it work because they weren't getting paid and they were doing it very slowly and badly But they were doing something that was useful.
54:02
[SPEAKER_01]: They were you know moving whether it's just moving objects about in some sort of assembly room or assembling tote packages together It's not yet fast enough reliable enough or it's not the right price yet, but they are doing that stuff and it is going to get cheaper
54:15
[SPEAKER_00]: I feel like on that point too, like when I think about the threat that immediately comes to mind to me, it's not like this kind of mass replacement of human workers with humanoid robot workers or robot workers in general, but like this story I remember from a few months ago where I think it was like a Japanese convenience store or something like that, you know, had a human worker wearing the types of things that you were talking about, you know, the gloves and stuff like that, doing the task to train the robot to be able to do the same
54:44
[SPEAKER_00]: that's the type of places where I feel like there's a potential risk if the technology can be made reliable enough right.
54:51
[SPEAKER_01]: Yeah and automation is always used as a threat to discipline labor and we've seen this with AI so many times and you know the people like McDonald's and these companies saying oh we're going to put a large language model in the order kiosk and that's going to take rid of it.
55:04
[SPEAKER_01]: And it's like, no, they fail, they fail, they're not very useful, they'll get better in time, but the main utility of that in the near term is saying, don't ask for more money and don't unionize and don't do this and don't do that.
55:17
[SPEAKER_00]: And we're going to see the same with human ones as well.
55:19
[SPEAKER_00]: I think that makes a lot of sense.
55:20
[SPEAKER_00]: And on that, and we kind of talked about this a bit earlier, but
55:23
[SPEAKER_00]: I feel like the desire or the motivation to create humanoid robots comes so much out of these science fiction visions that we've all encountered and that I would say many of the people leading the tech industry have been really kind of steeped in for many years and believe we should try to realize.
55:40
[SPEAKER_00]: Do you think this is that these ideas are harmful in?
55:44
[SPEAKER_00]: shaping how we believe the future should look and kind of trying to guide the development that we expect to see or at least that these CEOs are trying to undertake.
55:55
[SPEAKER_01]: I don't think they have to be totally harmful.
55:57
[SPEAKER_01]: I think, you know, some people may say I'm a little bit too optimistic or credulous even about some of these claims these companies are making, but I spoke to a lot of these executives and engineers and so many people had very hopeful
56:13
[SPEAKER_01]: I think there is a great potential for using humanoid robots and robots in general for stuff like elderly care, for helping out in medical institutions.
56:22
[SPEAKER_01]: I want to take the example of this Japanese toilet cleaning robot.
56:26
[SPEAKER_01]: I mentioned the very simple robotic form, but I think that's a good use of a robot.
56:32
[SPEAKER_01]: I think a human who was perhaps doing that job might be quite pleased if a robot was doing it and said if they were able to do other things with whatever ambitions they had in their life, I don't know what they would be.
56:43
[SPEAKER_01]: There is automation and technology that has already eliminated incredibly back-breaking jobs that were done by humans for thousands of years, for hundreds of years, and that is to humanity's good.
56:54
[SPEAKER_01]: The problem is the social, the political systems that then distribute the wealth that comes from this technology.
57:01
[SPEAKER_01]: Obviously, that is where we are falling down at the moment, but the technology itself has such a wonderful potential that we shouldn't dismiss it.
57:09
[SPEAKER_00]: I hear you on that.
57:10
[SPEAKER_00]: And I think that that is a worthy argument to make, right?
57:13
[SPEAKER_00]: Especially when we can see how technology has clearly been used to help people in the past.
57:18
[SPEAKER_00]: These technologies can be used to help people as well, even, you know, kind of broader automation technologies.
57:22
[SPEAKER_00]: But it's the question of, like, who controls it?
57:24
[SPEAKER_00]: Who benefits from it?
57:25
[SPEAKER_00]: And right now, if it's like the Elon Musk's that are benefiting from it, that's where the risk comes in, right?
57:30
[SPEAKER_01]: Yeah, that's the danger.
57:31
[SPEAKER_01]: I went to this robotics conference in London.
57:34
[SPEAKER_01]: And actually one of the most interesting talks I saw, again, got left on the editing room floor, was about someone who's trying to make open source versions of all these robots, who looks at these designs and says, look, I made my inversions of them, anyone can print them out.
57:47
[SPEAKER_01]: And here's point, you know, he was a complete robot nut, let's say, he just, you know, he loved the technology, he loved the potential, and he was so worried about who's going to get to control this stuff.
57:57
[SPEAKER_01]: But because, you know, okay, the US might be worried about China being able to 3D-print its own workforce or whatever that's like, but it's also very worrying if your Musk and 3D printers don't work for what automation does is it strengthens the hand of capital holders and it reduces the power of labor within the marketplace and so yeah.
58:17
[SPEAKER_01]: We need to deal with that politically first.
58:19
[SPEAKER_00]: Yeah, and to be quite honest, I think that that is a great place to leave this conversation, this kind of assessment of where the humanoid robots are at, and what we really need to be thinking about as the real problems here.
58:30
[SPEAKER_00]: James, really great to talk to you.
58:31
[SPEAKER_00]: I'm sure we'll be on the show in the future, and I'm looking forward to the work that you do next, but thanks so much for taking the time.
58:37
[SPEAKER_00]: I really appreciate it.
58:38
[SPEAKER_00]: No, thanks for having this conversation.
58:39
[SPEAKER_00]: I really enjoyed it,
58:43
[SPEAKER_00]: James Finson is a journalist, author of Beyond Measure, and a former senior editor at The Verge.
58:48
[SPEAKER_00]: Tech Won't Save Us is made in partnership with the nation magazine and is hosted by me, Paris Marks.
58:52
[SPEAKER_00]: Production is why Kylie Huesen, Tech Won't Save Us relies on the support of listeners like you to keep providing critical perspectives on the tech industry.
58:58
[SPEAKER_00]: You can join hundreds of other supporters by going to patreon.com slash Tech Won't Save Us and making a pledge of your own.
59:04
[SPEAKER_00]: Thanks for listening and make sure to come back next week.