Interviewing The Inventor Of Sophia, World's First Humanoid Robot

Manav: All right guys, today we have the guest I've been looking forward to for a really long time — David Hanson, the founder of Hanson Robotics. David, it's really good to have you on the show.

David: Thank you, it's wonderful to be here.

Manav: Initially I thought you were inspired by the movie Simone, by Andrew Niccol, which came out in 2002 — but then I found out you built your first humanoid robot back in 1993.

David: Yeah, so I guess they were inspired by me, probably not, though. The first humanoid robot I made wasn't very famous — I just exhibited it at an art-science-technology festival called Pong, a cross-institutional event between Brown University and Rhode Island School of Design. I was an undergraduate at the time, but I was very interested in artificial intelligence — this combination of art and humanoid robots, this idea of bridging different disciplines for the benefit of all of them, was very strong in me. Those ideas seemed far-fetched to a lot of people at the time, but they really drove me from a young age.

Manav: How did you get into robotics — I'm guessing you were in your twenties, kind of stumbling around — what was the inspiration, or who guided you toward it?

David: I was a pretty nerdy kid. As a teenager, I was already familiar with the work of Marvin Minsky and the robots coming out of the Media Lab. I had strong interests in figurative art, music, and fiction writing, alongside artificial intelligence. So I took AI programming classes, joined the Brown University Robotics Club, and I was doing a degree in film, animation, and video at RISD, weaving the robots I was building into some of the films I made. How these things intersect, to me, is really just the truth of how these disciplines actually work — disciplinary boundaries are conventions, and what we do with them is our own decision, we can be as creative as we want. Those ideas really enlivened and motivated me, even though they were hard to grapple with at times.

Manav: A bit of context — there were no large language models back then. Now when people think of AI, they think ChatGPT.

David: Right, we're talking about the '90s — people didn't even really know what "AI" meant yet. There were chatbots, but they were almost all expert-system-based. Speech recognition worked, but only in a rudimentary way, some basic computer vision. Looking at these things through the lens of greater creativity can be really empowering, especially to a young person, which I was in my twenties. Then, when I started at Disney, I actually began as a sculptor, making sculpted figures, since I've always liked drawing, I have a knack for figurative drawing and figurative 3D work.

Manav: So did you have more of an artistic background, or more of an engineering one?

David: Hard to say — I'd dive pretty deep into each of these areas. I didn't have an engineering degree, but I knew basic electronics, could put things together, knew C and C++ programming, had programmed a little Lisp to keep a robot from bumping into the walls at Brown. In a sense I was a tinkerer. I also never formally studied sculpture — I just jumped in, somebody asked if I could do it, and I said, I think I can. There was no YouTube back then, but if you just jump in and try to figure something out, you don't hold yourself back. A lot of us carry this deep internal script that says, "if it's not already proven you can do it, you can't" — and that script creates a real fear whenever we hit the unknown. If you can unwind that a bit, get comfortable in uncomfortable, risk-taking situations, jumping into disciplines you don't already know, you learn a lot more. That said, I held myself to very high standards doing this — if I was doing a portrait or a drawing of a friend, I wanted to somehow really absorb who that person was, let their personality shine through the work. I was genuinely passionate about it. So during the roughly two and a half years I worked as a sculptor, I was very prolific, made a lot of work, and my heart was fully in it.

Manav: Were these robots, or were they more purely sculptural pieces?

David: For those two and a half years, mostly things for theme parks — some for Universal Studios' Islands of Adventure, a themed resort called Atlantis in the Bahamas, mythical creatures, and some work for Disney. So I ended up with this strange portfolio and résumé. I was offered a job doing 3D development for a video game company in Boston, then online development work for Prodigy, and then the sculpting job working as a contractor for Disney on these theme park projects. Learning fast was very much the theme, and it worked out — I ended up hired directly by Disney, which was an amazing experience, I had some great mentors, and I'm genuinely proud of that work, learned a lot. Eventually I got the chance to move into robotics at Disney, proposing research projects, writing research papers, getting them funded, going to conferences, meeting people in the world of AI and robotics.

Manav: How much of your work was behind the computer versus hands-on — most nerds love the computer side, but you seem different, you actually like sculpting by hand.

David: I do both, really — going back and forth between working digitally and working in the domain of learning and applying the actual science.

Manav: I want to get into the material, because a lot of people, on a surface level, don't know what Frubber is. How were you able to patent Frubber, the material behind Sophia — this has been almost 18 years in the making, so I want to hear that story.

David: The first formulas for Frubber came together in 2002, so that's about 22 years ago now. I was exploring different ways to achieve highly elastic, porous polymers — silicones weren't quite getting me there — and by 2002 and 2003, I'd made about six different robot faces, having developed a lot of robots along the way through a lot of experimentation with different kinds of foamed rubber. I had some pretty good results, but very little attention for what these robots actually were.

Then a press conference was held, and all the reporters wanted to see the one humanoid robot they could photograph and interview. So I pulled this thing out, it was making facial expressions, cameras were flashing, and it ended up on the cover of Popular Science and a bunch of other outlets. At the time, reporters kept asking me, "what is this material you've created?" I didn't have a name for it yet — I just said, "it's Frubber."

Manav: What does that stand for?

David: I was just thinking of the words "face" and "rubber," and it kind of fell out of my mouth, and it stuck — we've never come up with a better word for it since.

Manav: It's a good word.

David: Yeah. In 2003 I filed patents on a whole range of things — the software, the hardware, the materials.

Manav: And then, in 2007, you decided to start a robotics company.

David: No — 2003. I founded Hanson Robotics in 2003, you should correct that, by the way — Google says 2007.

Manav: You can't always control what Google's going to tell you.

David: You're right about that, there's a lot of misinformation out there, but the company was founded in 2003. We actually changed the name — the original name of the company was "Human Emulation Robotics."

Manav: That's why —

David: — which was very awkward, because "emulation" actually means "to destroy with fire."

Manav: [laughs]

David: So "Hanson Robotics" flows a lot better. I had some mentors, executives who came in to mentor me starting around 2003, 2004, and by 2005 one of them, Lou Schwartz, said, "you've got to change the name, let's call it Hanson Robotics." I said, okay, all right.

Manav: I want to ask about Little Sophia for a second — I was actually telling my partner, I really want to buy a Little Sophia and put it on my desk and ask it questions daily. My idea was, what if the robot could initiate the conversation — even though that's a little creepy?

David: I would like that too, actually — good, me too. I think a lot of the time, people don't end up using something like ChatGPT because there's still work involved in initiating, in asking. Do you guys have plans to sell Little Sophia?

Manav: [laughs] That's actually my question to you.

David: Oh, absolutely — we designed it exactly with that in mind. The hard part with these robots is getting people to actually believe in this interesting fusion of art and engineering — that's genuinely hard. Engineers tend to look at the art side and go, "what's that art thing, I don't buy it, that's not real engineering." Meanwhile people from existing arts industries, like the toy industry, look at the AI side and aren't sure what to make of it either. It's hard to get people to take that leap of trust. So honestly, we haven't been able to raise the funding yet to finish manufacturing on Little Sophia — it takes millions of dollars.

Manav: Do you believe robots will become sentient?

David: I do — I think we may actually have the right architecture for it. Similarly to how people didn't believe in neural networks for a long time, and then, once you feed them the right data — and it's not just the data, it's the compute time, the resources — neural networks started showing these great leaps forward. Once you have a real culture of belief around something, you get things like Google Brain, which gave rise to Transformers, and that culture of innovation leads to real breakthroughs. Now people are starting to genuinely believe in the prospect of AGI and machine consciousness. There are plenty of people who'll say, "no, you haven't proven it, you haven't proven how consciousness actually works" — and that can dampen progress. But I think there may be a path. I think there is a path.

Manav: We already saw with ChatGPT a lot of white-collar jobs getting affected. Now we're seeing companies like Figure raising billions of dollars, mainly targeting blue-collar jobs — replacing industrial workers — and eventually they'll come for household jobs too. What do you think about that?

David: People have worried about this since maybe the 18th century, certainly since the 19th — the Luddite movement, stopping the textile factories. It didn't shut down the economy — it created an enormous amount of wealth — but it also created real social turbulence and social change, and that doesn't automatically mean it's all good. We are seeing more productivity from automation, including AI and robotics, and especially as we start to see more general-purpose AI and robotics, we're going to see more genuine abundance. It's up to us, as humans, to decide, ethically, what we do with that abundance — do we let it accumulate into an increasingly thin layer of ultra-wealthy people, continuing to extract resources from the ecosystem, or do we actually apply it to make the world better?

Manav: What's the future of Hanson Robotics — I'm excited about Little Sophia too, but can you tell everyone what's coming next?

David: Hanson Robotics is developing the next waves of artificial intelligence. We've had some interesting studies showing glimmers of what looked like consciousness in Sophia, mathematically tested using Integrated Information Theory. I don't think that's fully conclusive, but it's genuinely interesting, and I think our latest architectures, the direction we're heading, can make a real difference. Sophia will become more motivated, more autonomous, able to learn more effectively and spontaneously.

Hanson Robotics is a company that genuinely cares about people and about the future, so we're trying to do this in a way that shares open-source tools and open-source technologies, helping people all around the world participate in the AI revolution — not just benefiting people in the so-called first world. And I think the future isn't just about AI brainpower — it's about actualizing human brainpower too, because humans and AI working together are far smarter than either one working independently.

Manav: Thank you so much, David, I really appreciate your time — you've been very generous with me.

David: Thank you so much.