Thanks largely to advances in AI, robots are now capable of things never before possible: picking delicate vegetables, moving heavy goods deftly around a warehouse, making pizzas. But there are still limitations to even the most sophisticated and expensive machines. Replicating human dexterity, for example, remains a surprisingly fiendish challenge. While some experts talk about an imminent “ChatGPT moment” for robotics — one that has the potential to completely transform society — it’s not quite clear when this moment will arrive, or what it will mean when it does. In this episode, we explore how new robot technology is reshaping society, the economy and our understanding of what it means to be human.
Thanks largely to advances in AI, robots are now capable of things never before possible: picking delicate vegetables, moving heavy goods deftly around a warehouse, making pizzas. But there are still limitations to even the most sophisticated and expensive machines. Replicating human dexterity, for example, remains a surprisingly fiendish challenge. While some experts talk about an imminent “ChatGPT moment” for robotics — one that has the potential to completely transform society — it’s not quite clear when this moment will arrive, or what it will mean when it does. In this episode, we explore how new robot technology is reshaping society, the economy and our understanding of what it means to be human.
Featured in this episode:
Stefan Glibetic is the founder and CEO of Mycionics, which builds robotic harvesting solutions for the global mushroom industry.
Robert Brooks is the founder and CEO of ForceN, which manufactures joints, fingers, wrists and ankles for surgical, industrial and defence/aerospace robots.
Tarek Rahim is the co-founder and CEO of Mirsee Robotics, which specializes in building humanoid robots for various commercial sectors.
Nick Dechev is an associate professor in the faculty of engineering at the University of Victoria. He is also the founder of and chief technology officer at the Victoria Hand Project, a non-profit providing affordable prosthetic devices around the world.
Jason Millar is an associate professor in the faculty of engineering at the University of Ottawa. He is also the Canada Research Chair in ethical engineering of artificial intelligence and robotics and director of the Canadian Robotics and Artificial Intelligence Ethical Design Lab (CRAiEDL).
Further reading:
New research suggests consistency, not complexity, is the key to teaching robots dexterity
The $6 billion Chinese startup trying to build hands for every robot
How physical AI is reshaping robotics today — and what comes next
Ten ways humanoid robots are about to change everyday life
Beyond bionics: How the future of prosthetics is redefining humanity
Subscribe to Solve for X: Innovations to Change the World here. And below, find a transcript to “Degrees of freedom.”
Manjula Selvarajah: If you could have a robot take over something for you in your life, what would it be?
Conference attendee #1: Ooooh, that’s a big question.
Conference attendee #2: I love doing dishes, I love cooking, but, like, folding clothes is where I draw the line.
Conference attendee #3: Prepare my dinner.
Manjula Selvarajah: You can’t make your own dinner, no? Too much work?
Conference attendee #3: I can do it, and I do it, and I like doing it. But I’m very, very busy.
Conference attendee #2: Whenever people mention robots, it’s always just, like, this serious, very scary idea. But, like, I would like to see robots embracing whimsy of some sort. Maybe a useless robot. I would love a useless robot that just follows me all day and, like, makes jokes or something.
Manjula Selvarajah: (laughs) OK, I can tell you, I have not heard that.
Narration: I’m Manjula Selvarajah, and this is Solve for X: Innovations to Change the World. Earlier this year, I paid a visit to Discovery X. It’s a tech conference held every year in Toronto. You know the scene, a big assembly, tech reporters moving through the crowd with their microphones, and … robots.
Manjula Selvarajah: So it’s not, this is not teleoperation right here when it lifts the block up?
Tarek Rahim: No, this is pure embedded intelligence. It’s all running on the computer inside the robot. I think it’s safe to say, like, every industry has been transformed by AI. The next evolution of AI is physical AI, where you apply that exceptional intelligence into a physical body, and it does real physical work in the world.
Narration: There’s both excitement and fear about how fast this technology is advancing and how it’s reshaping society and the future of work. But even as we see AI agents get better at solving problems or writing their own code, getting AI to interact with the world around us, that’s still a fundamental challenge. And when it comes to the task conference goers talked about, cooking or folding laundry, that depends on something very human and almost deceptively obvious — our hands. Here's Ellie, producer of Solve For X.
Ellen Payne Smith: One of the things I have looked into and learned about is this idea that what could possibly be holding back robots from being useful or truly present in our everyday lives. They’re all around us. A washing machine before could be considered a robot in some ways. There’s automation in factories for, you know, the last 50 years. These things do exist, but what is holding back robots from being useful in our lives? I had done a little bit of light Googling and learnt that manual dexterity and replicating what a hand can do and just those precise movements is really difficult for robots.
And doing that in an environment that isn’t controlled, meaning that, you know, is variable like our homes, is also really difficult for robots. And that made me really interested, and I wanted to look more into it. And so I looked for someone who had been tackling this problem in a really interesting way.
Stefan Glibetic: My name is Stefan Glibetic, and I am currently the CEO of Mycyonics.
Ellen Payne Smith: What Stefan’s been trying to do is he's been trying to develop a robot that could help pick and harvest mushrooms.
Manjula Selvarajah: Interesting.
Ellen Payne Smith: So the story starts back when he was at university.
Stefan Glibetic: Yeah, so back in 2013, a mushroom farmer, Murray Good, approached the University of Western Ontario, where I was an undergraduate student studying mechatronics, which is essentially robotics. And, basically he was like, "Can modern technology help the mushroom industry? Because we don't have any modern technology."
Ellen Payne Smith: So he started trying to help this farmer.
Stefan Glibetic: Everything is done by hand mostly. It’s very labourious. It’s dangerous. It’s a difficult climate environment to be in for a human harvester all day long, and it’s just never-ending, back-breaking work.
Ellen Payne Smith: So Stefan decided to do some research, and he found that it’s been something that they’ve been trying to do for over 40 years.
Manjula Selvarajah: Wow.
Ellen Payne Smith: But also, there’s parts of a mushroom farm that are really tricky. So there’s, like, the robotics part of it, and then there’s the mushroom farm. Maybe I can just describe to you the way he described a mushroom farm.
Manjula Selvarajah: Yeah. I just imagined sort of a field of little, white little things popping up, but I don't think that’s what it is. What does that look like?
Ellen Payne Smith: It’s warm. It’s very humid, and it’s kind of dark. And he described the architecture of the farms, and there’s these beds, like sort of bunk beds, of mushrooms growing.
Stefan Glibetic: And there’s an ocean of mushrooms on these beds.
Ellen Payne Smith: But, I think we both know that, you know, machines don’t love moisture, so that has also been something. But think about a mushroom. Getting a robot to pick a mushroom is a really complicated task. How do you get a machine, an object made of metal, programmed, you know, working with electricity, how do you get that to grasp a mushroom that even with the slightest wrong movement then can bruise and go black and rot? Like, they're so delicate.
Stefan Glibetic: Sometimes they’re very easy. They pop off with just a little push. Sometimes they just fall off themselves. So you have all these different shapes and sizes and configurations of mushrooms, and the rate at which they grow is overwhelming.
Manjula Selvarajah: So there’s that issue. So you have the humidity. You have how fragile it is, and then you also have this variability. So, this is what makes a hand with a set of human eyes actually quite good at this.
Ellen Payne Smith: Right. So a large part of what Stefan has done and the success behind his work today is developing a robotic hand to achieve this. And he’s taken a lot of inspiration from the human hand, which is very cool.
Stefan Glibetic: The traditional automation type approach is to use suction cups for pick-and-place type of jobs. But the mushroom cap is super delicate. So we were like, “OK, why do humans do this really well?” Humans, they just use their fingers. And if you think about traditional robotics, typically robotic grippers are just another kind of stiff material. The whole soft gripper or soft finger is more of a modern approach to dealing with some of these problems, but a lot of grippers in the industrial space are designed like that. They’re very stiff, they're strong, they’re designed to very precisely grab a specific type of also fabricated material that was designed to be also picked up by a robotic gripper of some kind. But obviously mushrooms are not.
Ellen Payne Smith: In the pilot that you have had and, you know, in terms of the deployment so far, how much better is a mushroom harvester? Is it at the level of humans or is it exceeding humans?
Stefan Glibetic: It was matching human harvesting quality and in many cases exceeding human harvesting quality because we only touched the mushroom in a very particular spot. The issue for us was that we just didn’t have enough arms. The growth was just too much for us, so that’s what we’re actually deploying next month here at the farm, we’re deploying four robotic arms to pick mushrooms. It’s a hybrid solution, so we have robots and people working together. That’s kind of like the cobot, more modern automation approach is where people do the more delicate, highly dexterous, complicated type of tasks that would just make robotics too expensive and too delicate to operate industrially and reliably while the robots do like the brunt of that back-breaking work.
Manjula Selvarajah: So when you went in with this question about dexterity, you know, one of the big answers that you got is that he is inspired by the human hand.
Stefan Glibetic: From a robotics perspective, it’s something unimaginably difficult. I don’t know whether we’re going to ever be able to achieve something like that even in my lifetime. Each hand can move in 25 different ways, their so-called degrees of freedom. Each degree of freedom represents one very particular type of motion, like going up and down or left and right. Robots are typically in the less than 10 degrees of freedom realm, and and I'm talking state-of-the-art. It’s just unbelievably sophisticated and awesome.
Ellen Payne Smith: I went in thinking, oh, it’s all about the mechanics of the hand, and he really helped me understand just how important the computer vision is.
Manjula Selvarajah: The other thing I was thinking about is, is touch and sensation.
Ellen Payne Smith: The question of could you create a more dexterous robot hand that actually has sensation? I mean, that’s fascinating. I’d love to learn more about that.
Narration: Stefan really underscored what a marvel of engineering the human hand is, and that’s just talking about the way the hand moves. What about the way the hand touches, senses, and feels the world? How close are we to getting machines to do that?
Robert Brooks: I’m Robert Brooks. I'm the founder and CEO of ForceN. We build superhuman manipulation for robots and for physical AI, and I’m here at our headquarters in Toronto
Narration: Nature spent millions of years designing the hand. Engineers like Robert are now trying to build something comparable in just a few decades.
Robert Brooks: And that isn’t even touching, trying to replicate all of the touch and sensation in the hand is another whole problem that we’re just starting to work on.
Manjula Selvarajah: Can you actually give me an example of something that we do often, and kind of walk me through it, that explains the kind of the power, the dexterity of the human hand?
Robert Brooks: Yeah, so the human hand has a few different ways that it feels and interacts with things. I think probably the best example is if you want to find something in your backpack or purse, you just put your hand in there, and you rummage around, and you feel, and you characterize something, and you’re like, “Hmm, this is a pen.” There's a lot of things that go into “Hey, it’s a pen.” And, you know, maybe it’s not quite right, you end up picking out a mechanical pencil or something, but you’re really, really good at that, and that’s something that robots can’t even begin to do right now, is picking things out of a messy pile.
Narration: Robert explained that e-commerce companies would love robots that can pick through things reliably. Just think of the sheer variety of everything you scroll through online. And despite companies like Amazon driving automation forward, much of that picking is still done by people. But how do you translate sensation, something that lives in the body, a kind of biological intelligence, into a signal that a machine can read?
One early breakthrough came out of MIT in 2009. It’s called GelSight. It turns touch into something closer to an image recognition problem. Using a soft gel and cameras, it reads texture, not by feeling it, but by seeing how it deforms. Now, this is just one approach to touch. Others are working on artificial skin. Robert, meanwhile, is approaching the problem from a different direction — pressure. Specifically, force and torque.
Robert Brooks: So don’t just think about force statically, think of it dynamically, like over time. So is something squishy? If I tap it, does it vibrate? Is it hard? We also measure temperature. So those are all the things that we use to kind of understand what it is we’re grabbing, and then that determines how we interact with something.
Narration: The origins of this technology trace back to Robert’s PhD work at the University of Toronto when he was looking at how robotics was revolutionizing surgery.
Robert Brooks: So we started out actually doing surgical robots. We have a huge number of ways of imaging the body, but surgeons have largely lost access to the sense of touch they had in open surgery. So the goal is not only to recreate that sense of touch, and then on top of that, we can start to augment what human touch is capable of doing. So you get — the surgeon gets their full sense of touch, and they get to have their minimally invasive surgery, too. Think of this like a microscope for the hands.
Manjula Selvarajah: So the sense is you’re allowing surgeons to feel things without resorting to open surgery.
Robert Brooks: Exactly. You have the ability to amplify human touch so you can, for example, feel the inside of a blood vessel. We’re working with a French company. They’re working to build a neurovascular, endovascular catheter robot, and their system is far more sensitive than the human hand, so they can start to feel things that are just at the edge or beyond human sensitivity perception. Kind of like you can use a microscope to see things that you can't with the naked eye.
Manjula Selvarajah: You know, why is it that, you know, they can do complex tasks like surgery, but they still struggle to do a simple thing like folding a shirt?
Robert Brooks: Yeah, so you can get, uh, if you take a surgical robot, you can fold shirts really well. Part of it is a surgical robot has a lot of sensing capability. These are million-dollar machines. These are, you know, the F1 cars of the robotic world, and, you know, you’ve got a human on the other end.
Manjula Selvarajah: So it’s just expensive.
Robert Brooks: Part, part-
Manjula Selvarajah: It's expensive and it's enabled by a human. Is that what it is?
Robert Brooks: Yes. So now we’re trying to get AI to do it. So the AI has to learn to do it, and we’re trying to build humanoid robots that come close to what humans can do. And it’s the same companies building all the components for these early humanoid robots that built the components for surgery. And in a lot of cases, it’s almost the exact same components. So we’ve really come down the cost curve and gotten very sophisticated with image sensors. Now we need to do a lot of the same things for touch in order for this to become ubiquitous so that it can do dishes, and there’s gonna be a lot of work in between.
Narration: But touch goes much further than handling objects. Researchers believe that replicating it could help spark the development of more capable autonomous systems and even change how machines learn.
Robert Brooks: You know, Holy Grail is being able to one-shot a task badly. Because that’s what humans are great at. You can put a human in front of just about anything, and they can do it badly the first time, but they’ll do it.
Narration: Robert tells me his sensing technology is going to be integrated into humanoids soon for a company called Agility based in Oregon. They call the robots Digit. Interestingly, the hands don’t have fingers at all, no digits, at least not on the versions being shipped to factories so far. They’re more like claws or paddles built for picking up boxes and moving totes around. Earlier this year, Agility announced a deal to put seven of its robots to work at a Toyota plant in Ontario. Back at the Discovery X conference, I found another company with a similar pitch, a humanoid robot made in Canada.
Tarek Rahim: So I’m Tarek Rahim. I’m one of the co-founders of Mirsee Robotics.
Narration: Tarek’s humanoid looks a lot like what you’d expect. It’s made of metal, about my height. Instead of legs, they’ve opted for a wheeled base. Look closer, and it appears the most detailed engineering is in its fingers.
Manjula Selvarajah: I’m looking at the hand there, and it looks like a hand, like a kind of a metal version of a hand. Talk to me about how difficult it is to replicate a hand and, and whether you need to, actually.
Tarek Rahim: If you want to build a robot that can do the widest number of tasks, you really do need a human hand. You can get away with fewer fingers for some applications, but there’s a reason we evolved five fingers. That’s why most humanoids are going in that direction of delivering human-like form.
Manjula Selvarajah: Like, I’m seeing it reaching for a little red block on that tray in front and moving it into another spot. And it does it so delicately, right?
Tarek Rahim: And it’s semi-random where that block rolls off onto the table. So we’re demonstrating that you don’t need highly structured environments. These robots are actually reasoning and determining. These aren’t hard-coded animations.
Manjula Selvarajah: So when I look at something like this, I mean, pardon me for saying this, but it looks like a simple movement. So tell me —
Tarek Rahim: It looks very simple, but it is immensely complex. Like, the —
Manjula Selvarajah: Why? Why is it hard? Why is it so hard to replicate that? Like, this is something that a two-year-old could do.
Tarek Rahim: Because there’s a lot of complex vision that’s required to do this. You need spatial intelligence. You need to do edge detection. You need to do object classification. Humans are, you know, very smart in a lot of things that we take for granted. When you build a robot, you have to understand all these tasks and processes and replicate them, and that’s why it is so immensely complicated, and that’s why the hand just missed the block. Because it was off by just a millimeter, and that was enough for it to drop. We take motion for granted as humans, but there is a huge portion of our brain just devoted to moving our muscles around and navigating through space and spatial awareness.
Manjula Selvarajah: And where do you see the first point that this will enter, you think?
Tarek Rahim: The one industry that is the most easily serviceable is manufacturing because what they want can be delivered. They don’t want the complex features like the consumer market, for example, wants. It’s going to be probably a decade before these are realistically in our homes doing laundry, cooking our meals, and that’s just because the randomness and the expectations of the everyday person is much higher than, you know, what a business wants.
Narration: I find it fascinating how something as intuitive as grasping, feeling, or picking up a block can be so fiendishly difficult to replicate. There’s actually a name for this, Moravec's Paradox. It’s named after Hans Moravec, an Austrian born Canadian computer scientist. In 1988, he wrote, “It is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility.”
All these years later, this holds. And nowhere does solving the problem of dexterity matter quite so much as in the world of biomedical design and prosthetics.
Nick Dechev: My name is Nick Dechev. I’m an associate professor of mechanical engineering at University of Victoria. I’m also chief technology officer and founder of the Victoria Hand Project, where we do prosthetic limbs.
Narration: The Victoria Hand Project provides prosthetic hands to people in need around the globe, most recently helping Palestinian refugees in Egypt.
Nick Dechev: So the Victoria Hand is a combination of 3D-printed and steel components. The primary motivation is high function, high reliability, and very, very low cost.
Narration: Which brings us to the question Nick has spent his career chasing.
Nick Dechev: One of my most sought after questions is how can you get the intention of a human being translated into the mechanical action of a prosthesis? And it’s a pretty difficult question because there’s a lot going on between when you’re first forming a thought to reach out and grab something and actually having a machine do it.
Narration: Nick shared a statistic that came as a surprise. About 50 percent of people will stop using their prosthesis within the first six months.
Nick Dechev: Could be a $50,000 or $100,000 prosthesis. It doesn't matter. There is a lot going on in the minds of a person, both in terms of the physical need and their emotional needs. When the outcome is successful, it’s a great feeling because I’m walking away thinking hopefully this will really make their life a little bit easier and a little bit better. And that’s really important to me.
Manjula Selvarajah: You know, in doing the research that we did for this episode, one of the things that I’ve noticed is it seems like robotics may have hit this kind of inflection point where you see things that are happening when it comes to soft touch, computer vision, sensors, and of course, AI. What excites you the most?
Nick Dechev: It’s 100 percent, it’s the AI. To have a machine operating in the world, it needs to be context aware, and that has been missing for the entirety of robotics up until now. So the computer vision that you mentioned, that problem has largely been solved since I would say the very early 2000s. There was an incredible amount of effort because a camera can gather all that information around the environment and bring it in. But then what we were always missing is what does it mean to the machine? And now we’re on the dawn of it having a bit of context and making leaps. Like, I don't want to say cognitive because I hate to put human traits onto machines, because they’re just machines, but having that context to say, “Hey, what should I do next?” As opposed to, “What do I do?”
Manjula Selvarajah: Do you see these advancements transforming the design of prosthetics?
Nick Dechev: Hopefully, yes. I mean, I think prosthetic development has always been looking over at robotics to see what can be applied. So I think probably the most useful examples would be new types of actuators, and I would say sensors as well, what novel types of sensors are available to understand what the body wants to do, whether it’s the mind, the nerves, the muscles, and then try to translate that user intention into actionable control of the prosthesis.
Manjula Selvarajah: Sensing intent.
Nick Dechev: Yes.
Manjula Selvarajah: That’s what’s interesting to you.
Nick Dechev: Yes.
Manjula Selvarajah: Nick’s own research looks at ultrasound, using it to read tendons that remain after an amputation. Others are looking at myoelectric signals and BCI, which we covered earlier this season. While Nick is hopeful that the advances we’re seeing in robot dexterity could spill over and help people, he’s still clear-eyed. Because the economics here are just really tricky.
Nick Dechev: There’s very few people to pay for it. Even if insurance is paying for it, there’s not a lot of revenue. And so the development in prosthesis has been in fits and starts, and it’s a structural problem. Unlike a car there's millions of users, and you can pour money in, you'll get your return on investment. But in prosthesis, that's not the case. And so I think right now, AI, for the time being, is very low cost to implement, and that’s because it’s software. Once the software is created, once the machines are programmed, very inexpensive to use. So the question for me is, how can some of that technology be applied to prosthesis? And I’m not sure yet. Obviously, a lot of these AI models are text-driven, so that’s not really an input for prosthesis at the moment. And a lot of other robotics are vision-driven, and prosthesis are touch. And so yeah, I’m optimistic, to be frank, because, you know, people are very clever, and I’m sure someone is going to think of some very interesting use case.
Maybe people have to start wearing headsets, these augmented reality glasses. Maybe that gives the AI enough context to anticipate, oh, you know, the person's putting their hand right next to that coffee cup. I bet they want to grab that coffee cup, and I bet they don't want to drop it. And I bet they want to keep it oriented in this position so the water doesn’t fall out. That's that context I’m talking about that just does not exist right now in control of prosthesis.
Narration: That idea is exciting. A prosthetic hand that doesn’t just respond, it anticipates. Several companies are already chasing how to incorporate AI to do exactly that. Talking to Nick, what stood out to me was how centered his work is on the person, building technology to fit their life and needs. It stands in sharp contrast to how many describe the arrival of AI. And with some industry leaders now talking about a ChatGPT moment for robots being just around the corner, it raises real questions about what that could mean for society.
To help me unpack this, I spoke with Jason Millar, a robot ethicist, philosopher, and associate professor at the University of Ottawa.
Jason MIllar: It depends what we mean by the ChatGPT moment. And so that’s just, you know — I guess I’m hedging a little bit. And certainly there are some robotic systems, say, the ones that are operating in Amazon warehouses or there’s a robot that I saw at Boston Dynamics down in the US that’s really good at loading and unloading and stacking boxes, right? Helping to unload cargo from trucks or ship containers or whatever. And those technologies are really advancing rapidly. So if we’re talking about a ChatGPT moment for the Amazon warehouse robots, we’re probably already in the middle of it. If we’re talking about humanoid robotics, probably a little far off.
Narration: Much of Jason’s work has focused on technologies like self-driving cars, more recently agentic AI. But I wanted his perspective on the push we’re seeing for more dexterous robots, and what questions he sees this raising.
Jason MIllar: The hope and the vision and the sales pitch is always that it will make life easier. Robots are meant to do the dull, dirty, dangerous work. I'm sure you've heard that before in researching this. It’s a very common —
Manjula Selvarajah: Yes, and part of me, part of me, I’ll admit it, kind of wants a little bit of that.
Jason MIllar: Sure.
Manjula Selvarajah: Yeah. Like, if something’s going to do my dishes, something’s going to do some of the tedious stuff, yeah.
Jason MIllar: And, you know, if you look at the — if you’ve studied this and you look at the advertising from 100 years ago, it was always, hey, let’s get these machines into the home — the washing machines, the dishwashers. It was always like, we’re going to make women’s lives easier. They’re in the home, and, you know, it’s going to free up all their time, and they’ll just be able to have lives of leisure around the house. I mean, nothing could be further from the truth. I mean, we hear the same with — we heard the same things with word processors and spreadsheets. You know, the more you automate the tedious daily tasks in the office, you know, we'll get down to a seven-hour workweek because all the things that used to take all this time, we don't have to do it anymore. And that’s, like, certainly not the case, right?
If you talk to people, they’ll say, you know, they feel constantly overworked. They don’t have time to get their work done. Every software solution is just downloading more and more tasks to them that somebody else used to do. It’s like you’re kind of concentrating the work in one person instead of distributing it over a larger group of people.
So as the person who studies these types of technologies in a social context, that’s the kind of question that I ask. Are we talking about a future here where we’re unlocking this kind of magical future of, like, leisure and time spent doing things that we love? And for what benefit? Like, who benefits from this? Those are really the questions that I ask.
Manjula Selvarajah: Do you think there are benefits, though?
Jason MIllar: Oh, certainly.
Manjula Selvarajah: What could that look like?
Jason Millar: There are dull, dirty, dangerous jobs that people should probably not be doing, right? Including many of those types of warehouse jobs that, you know, the human body is frail. We don’t want to be subjecting people to stress and having them injured and having them harmed and things like that. So undoubtedly there are benefits.
Manjula Selvarajah: So when I think about our lives now and our lives going into the future, I feel like parts of us are already enabled by virtual robots, and soon parts of our lives, if not already to some degree, will be interacting more with physical robots. How then do you think we maintain our humanity?
Jason MIllar: Oh, I think we need to make very clear decisions, both sort of, you know, as individuals and as a society, so through creating rules, like rule-making, laws, regulations, that you know, we need to decide where we stand in the hierarchy here, right? What are we willing to give up to technologies like robots and AI, and what do we just want to maintain as being uniquely human? We need to have conversations about where these technologies fit in society and have those collective conversations. I think we are about to have a conversation of that sort with artificial intelligence that I don’t know that the companies were expecting.
So when I look at the latest public polls on people’s attitudes towards AI, it has so radically shifted in the last couple of years from largely ambivalent to completely negative. I mean, people really don’t like AI right now. I think that is largely because of the way that the technology has just been unleashed on society with it seems like very little consideration of what the impacts of that technology would be.
Narration: The public debate about the future of automation is ongoing, and no doubt will get more complex as we see robots reach new heights of dexterity. But inside robotics labs, a quieter debate is unfolding about where engineers should look for inspiration. What if replicating what the hand can do means moving away from its rigid form entirely? Back to Nick, professor of mechanical engineering at the University of Victoria.
Nick Dechev: There's something called soft robotics that has been developing over the I'd say the last 10 to 15 years, very much more so in the last five or six years, where there’s novel types of actuation.
Narration: Some of the most exciting work in that sphere looks to nature for inspiration, like octopuses, starfish, creatures that move without rigid joints.
Nick Dechev: In prosthesis, you need things to be super lightweight and use very little energy because you’ve got to carry around your power with you if it’s going to be an electric prosthesis. And so I think those advancements are really exciting because you can, you know, there’s journal papers that are written by researchers. You know, we can read it, we can email them and just say, "Hey, you did this really cool thing, and I’m working on this, and can I learn more, or, or can we work together?" And, you know, I think that that is alive and healthy in research and academia.
Manjula Selvarajah: But what does sidestepping anatomy in search of greater dexterity mean for people who need limbs?
Nick Dechev: It’s a very deep question. If you could literally be a super person and do this incredible thing if you had an octopus appendage, would you use it? And it comes down to, you know, preference and choice.
Narration: We’ve spent this episode looking at the building blocks of machine dexterity. And even as engineers keep reaching and robots get more capable, there’s also the fact that science is uncovering new things about ourselves. The biology of touch, the complexity of the hand-brain connection, and what we thought we understood turns out to have infinite layers. In that way, the bar for matching our biology keeps getting raised. We still have one over the machines. But I had one last question for our robot philosopher.
Manjula Selvarajah: Jason, are you a robot?
Jason MIllar: I hope not (laughs).
Manjula Selvarajah: How would I know? How do you know for sure?
Jason MIllar: As a philosopher, I’ve thought a lot about this, as you can imagine. You know, these are some of — these are, I mean, you could have a whole episode just on this question (laughs). How do you know you're not a robot? And people have written great, great papers about, you know, are we living in a simulation, and, this all just the matrix? Here’s what I’ll say. If you’re wondering whether I’m a robot, we've already lost the battle (laughs). So, this whole conversation is moot. I trust you’re not a robot. I trust that I know a robot when I see one, and maybe that’s what we should keep it. We should keep it that way, right? That’s one of these questions that I’m asking us to seriously consider and make the rules such that there is never a day where I’ m sitting around wondering, you know, "Is Manjula a robot?"
Narration: Solve for X is brought to you by MaRS. This episode was produced by Ellen Payne Smith. Jason McBride is our senior editor. Lara Torvi, Sana Maqbool, and Sarah Liss are the associate producers. Max Swain composed the theme song and all the music in this episode. Gab Harpelle is our mix engineer. Katherine Hayward is our executive producer. I'm your host, Manjula Selvarajah.