Societal Impact and Future Integration

This is the Guardian . Last month at the London Marathon, Sebastian S ua y smashed the sub two-hour barrier, shaving 65 seconds off the previous fastest time. 930, absolutely incredible. I've never seen anything like that . But it wasn't the only record-breaking race that caught the world's attention in April. It was human versus humanoid. More than a hundred robots racing against runners Sunday in Beijing . In Beijing's half marathon, a robot named Lightning beat the human record by a whopping seven minutes . It's just the latest in what feels like an acceleration of robotic breakthroughs powered by AI. And the country racing ahead of all the others? Lightning's birthplace. China . But beyond running really fast, what do we actually want from robots? What will it take for them to get there? And are we ready for what happens when they do ? I'm the Guardian Science Editor, Ian Sample, and this is Science Weekly Amy Hawkins you're the Guardian's senior China correspondent and you've been looking into China's robo tics boom. So first of all, who is this company behind the half marathon robot? Have they been in business for years or are they new? What's their story? So they're a company called Honor, which is actually a smartphone company , which has only recently within the last year or so pivoted into robotics and embodied AI, which is what people are calling robotics. I think that shows just how fast the sector is growing in China. Like it's not even one of the big, more famous robotic companies, but so many companies are trying to get in on this moment. China's got a much more sophisticated and low-cost manufacturing supply chain. So the parts of the supply chain that are needed to build these robots is very readily accessible in China and it gives China a bit of an edge in that part of the AI race. So this is really coming about through this sort of marriage of AI with robotics. I mean who are the most important companies? And are they the sort of the longtime big players or are they disruptive companies coming through? No, I would say most of the kind of big players in robotics at the moment are ones that have been founded, you know, maybe within the past few years. And actually, some of the bigger tech giants in China, like Baidu and Huawei, have struggled a bit to kind of pivot their energies towards robotics. But the big ones at the moment are companies like UniTree. They're the ones that made the dancing humanoid robots which have been on display at China's Springgala . You know, they do kind of amazing martial arts and dancing and they're probably the the leader of the pack and they plan to IPO this year and they want to raise more than six hundred million dollars and that would be China's biggest tech listing in a number of years. You talk about unit tree trying to raise this six hundred million at IPO. What is the scale of the investment going into robotics in China? It's really, really huge, in part because the government has decided to make it a priority and China has a very top-down economic model. And so, you know, where the government goes, money, invest ment, political will follows. And so the Chinese government has already launched what they call a national venture capital guidance fund, and that will invest one trillion yuan over the next 20 years, which is more than 100 billion pounds purely in robotics. And then many cities have launched their own funds on top of this. So the kind of regional governments are competing to attract the best companies. So Beijing and Guangzhou and South China have each launched funds worth more than one billion dollars to support robotics researchers and designers. And the market itself reached an estimated forty-seven billion dollars in twenty twenty four and it's growing at a rapid rate. It's felt like for decades that robots are really not progressed beyond this sort of, you know, almost like you're dustbin on wheels type robot. You know, you're very basic type robot. And it seemed like robotics certainly hasn't had its chat GPT moment where suddenly you get a total inflection and the technology is suddenly dramatically more effective and you know finds more markets. I'm interested in the kinds of robots that China is actually focusing on building. I mean are they going for a particular style of robot or are they looking to put robots into particular roles that are are ones that they haven't been doing before? Yeah you're right. I mean they kind of big question with these humanoid robots that we've seen, you know, winning races and doing kung fu dances and so on is kind of why? Like why would you need a robot that can run fast or do an impressive dance and where they're going to be more useful is in factories and in household settings where they can do things like household care or cleaning and cooking. And that is what a lot of Chinese companies are focusing on now is how to actually create robots that can intelligently move through space without having to be pre-programmed to do a certain routine. You know, for example, running a race is basically running in a straight line. It's not that difficult to ask a robot to do that , whereas putting a robot in an unfamiliar room and asking them to clean it is a lot more of a challenge, and lots of companies are focusing on that now. Amy, these companies have clearly got challenges on their hands with these robots. I mean how are they trying to tackle for example this issue of a robot having to to navigate a space that it's it's completely new to, it hasn't been in before. If you think about large language models which we're all familiar with, like chatbots, they're trained on a data set which is kind of almost unlimited of all the text on the internet and they can have so much data to train their models on. But for robots which are trying to figure out how to move through space, it's a lot harder for them to acquire the data that's necessary to train sophisticated models. There's a company that are approaching this in a few different ways. I spoke to a company recently that he's invented a glove that a human being can wear to do various tasks like cooking and cleaning and peeling a banana and folding laundry. And this glove is fitted with sensors which can then harvest that data and that glove can then be put onto a humanoid robot hand and the humanoid can learn from the data in the glove how to do certain tasks. That's one way. Another way is some companies are paying human beings to wear kind of GoPro cameras on our heads while they do daily tasks and they film all this data, and then that is fed into what's called these world models. And some companies are building 3D worlds in which to train their robots about how to navigate space. So there's a few different approaches and you know companies are each taking a bet on why their approach is gonna work Robots are making massive strides, but what will it take to get them from running and dancing to washing our dishes and fixing our sinks. According to many roboticists, it comes down to just one thing. It's the hands that are the critical thing. They're the things that come into interaction with the environment and do the work. Nathan Lapore is Professor of Robotics and AI at the University of Bristol and his focus is exactly what many of these companies are grappling with. How to give robots what he calls human-like dexterity. He told me where the field has got to so far. Years ago, it seemed like this kind of distant dream. And you couldn't really see how it was going to happen in the short term. But things are moving so fast in robotics now with AI, new hardware, new fabrication methods. We're getting close. It's kind of becoming clear, you know, how this problem will be solved, but we're still not there to actually get it, you know, fully to like the human levels of dexterity. Nathan, talk me through what's difficult about getting a robot to do what our hands can do? I mean, you know, you can think of these revolutions. We had the industrial revolution where we brought in machining to do a load of tasks that humans did, and then with computing, with AI we're thinking about having the sort of cognitive revolution where a lot of our cognitive powers are taken over by machines. But with the industrial revolution we never nailed this dexterous capability that you would want from a robot . What why has that been so hard? We we can make machines do very repetitive things, you know, and that that's how a lot of industry works. You know, you you've got a machine that's kind of tailor made for a specific job, you know, screwing caps on bottles. Whereas our hand, it's like a general purpose machine. The most obvious way is to put a motor in every joint of the hand. That's not how the human hand works. We have the you know tendons , muscles in our forearms, we coordinate within our body how the hand moves, and it's a really challenging design problem. Just crack that problem in terms of the hardware and the AI, and then to simplify it down so that it's you know can be made more easily and is more affordable. It's really hard. So it seems from the outside, from a sort of lathew that that robotics hasn't quite had its sort of chat GPT moment yet. And I wonder you know, are there say three big challenges that robotics needs to crack in order for it to get there when we see this kind of step change that we've seen with the LLMs with the chatbots. With robots, there's multiple things that need to be got right before they become really practically useful. And so like three things would be you know the body, design of the body, how it's actuated, you know, if you like the design of the robot hands, the brain, you need the AI, but also you need an AI that's appropriate for controlling robot hands and robots and dexterity. And that can be quite different. It's kind of a form of embodied AI. And then you need the sensing as well. Sensing is critical and probably the hardest part actually. You know, something that's the sophistication of like human skin. You can think of our skin as like high resolution sensor, you know, covering our hand that can feel the you know forces in in fine detail. You know, think about the kind of exquisite detail that you can feel with your fingers. Now to design that kind of skin for a robot it's really, really challenging problem and it's been basically holding the feel back for decades. But recently there's been an explosion of new companies building robot hands because of a need for humanoid robotics. And that is driving forward developments in miniaturizing motors, high power density motors, you know, all the all these other kind of constituent technologies that are needed in order to bring that together to make a hand . Nathan, a lot of the discussions around AI today will raise concerns around what it means for job losses. And one of the reasons is that a lot of the tech firms, a lot of big tech, talks about how AI is going to drive a lot of people out of their jobs , okay? Don't you ever sort of get worried about the disruption that might come when robots have the kind of physical capabilities that that that humans have? And so this will start hitting some of that labour, some of those jobs, because at the moment people say, well look, if you want to survive, you know, the AI revolution, what you need to be is like a plumber or an electrician. I don't worry too much about this because you know the evidence in the past is always when when one type of job stops, you know, they kind of work gets displaced into other valuable activity. There are other concerns of what will be done with those robots as well. You know, what what's their use within society? So humanoid robots with with hands, you know, they can be very useful as assistance for us. But you know, on the other hands, you know, they can be used as soldiers. They could be used, you know, in security. It's a powerful technology , and I think we'll need to see how that develops, and probably some controls will need to be put in place by governments as these technologies come through. I was wondering if we will all end up getting jobs repairing robots, but then I thought, well no, there'll probably be robots to do that, won't there? Are you absolutely right? The big change to come is when robots will be building other robots. Once that happens, the cost of a robot should come down because you won't need human robots at the moment all robots are built by humans. Sometime in the future robots will become dexterous enough that they will be rebuilding each other and improving those builds as well. That can be done with AI. You know, on the kind of self-sustaining loop where the robots will will become more dexterous. Coming up. China's robots are racing ahead. Could humans keep up? Amy, China is obvious ly leading the way in this sector. Where does the government see all of this going? Where will these robots fit into society? China does already have what some people call dark factories, so they're entirely staffed by rob ots and so you don't need to turn the lights on. And those will be factories in which the tasks are very kind of repetitive, simple tasks where they've decided to use robots rather than human workers, but for sure kind of part of the push towards developing more and more sophisticated humanoid robots is to get them to a place where they could replace human workers in jobs where they can't currently replace them. You know, what the government's really worried about is actually got a shrinking workforce because of the aging population. So, you know, it's worried about not having enough workers to work in factories and on the manufacturing line and also to look after its ever-increasing elderly population. But that says, you know, in industries where there are risk of people losing their jobs, for example, taxi drivers, there has been some pushback and from taxi drivers in Wuhan, which is a city which has seen quite a lot of robotaxis or autonomous taxis rolled out recently protested about the fact that you know these drop driverless taxis were taking their jobs. Actors have kind of complained about AI taking their jobs. And so there is a bit of a pushback now about the risk of job losses. But the view from the government is definitely very much like these robots solve a problem rather than create one. And when you're out and about in in China, do you do you encounter any in the real world? Yeah, when it comes to the humanoid robots, I would say that you mainly see them at press conferences and at kind of PR events. But when you're walking out in the streets, y you know, I haven't encountered a humanoid in the street just yet. But you do see some robot dogs, which is another kind of side of robot development which people get as pets, as well as that you know, some areas already have quite sophisticated drone delivery systems. So in Shenzhen, which is a city in South China, there are lots of places where you can order food delivery and then it will be delivered to you by a drone flying through the sky. And so yeah, again, I mean delivery and logistics is a huge industry in China and that's another field in which people are trying to develop robots to be able to operate in that space. But actually food delivery is one of the areas where human workers feel the most safe because they haven't yet found a robot who can run up ten flights of stairs to someone's office building or you know navigate a lift successfully. And so there are these kind of spaces in which robots aren't very useful yet . Must be fascinating to be over there and to watch it all, Amy. Thank you so much. Thanks for having me Thanks to Amy Hawkins and Professor Nathan Lapora. You can keep up with all of Amy's reporting from China at thegardian.com . This episode was sound designed by Joel Cox and the executive producer was Ellie Bury. We'll be back on Tuesday. See you then . This is the Guardian