Dream Machines: Physical AI, Robotics, Being Human
00:09:48 · Evan Helda
Ad[00:09:48] S1: This episode is sponsored by Bifrost, the simulation and evaluation platform for physical AI. In robotics, you don't really know if a model works until you know where it breaks. And the real world will not hand you those moments on demand. The rare object, the strange lighting, the one in 10,000 failure that only shows up when it matters most. That's what simulation is for. Not just generating pretty synthetic images, but making reality testable. The problem is, as most of us know, simulation is incredibly hard. You need a team of 3D experts, custom tooling, and weeks to months of setup before a team can run a single serious evaluation. Bifrost changes that. They make it easy for developers to build and scale high-fidelity simulations, drop their models into the exact scenarios they need, and evaluate against the edge cases before anything reaches the field. And the synthetic data, it comes with it. Perfectly labeled. Because in physical AI, the winner will not be the teams with the most data. It'll be the teams understand where their models fail and fixes them before they fail in the real world. If you're building robots and want to trust what you ship, go to Bifrost.ai. Yeah, I think the word is evolution. This is going to force us to think, act, and behave differently. And I think there is going to be this era where we don't realize how much we have to change in that way in order to thrive in this new era. And so it's going to be up to parents and modern day schools to really rethink what is taught. Uh, hard skills or even soft skills. It's almost inner world skills, right? To see the world a different way. And then alongside that, we have to build these tools and AI tools to encourage, Hey, yeah, talk to me first, says the AI and we'll get our head around things, but it has to then encourage some sort of action in the real world, I think for the effect to come full circle. And so I think that'll be the best type of tools. Is ones that bring you in to the power of the tech and then expel you back out to your family, to your friends, to the forest. I don't know. All kinds of ways to put you back into nature. I guess the word maybe to sum up that is we need tech to have wisdom.
00:12:07 · Lara Sophie Bothur, Evan Helda
[00:12:07] S0: Yeah. [00:12:08] S1: Technics of wisdom and wisdom is going to be rooted in understanding that we need connection with each other. And understanding people, I think that's something that you actually studied. You studied psychology. Yeah. Tech psychology. So give me the academic definition of tech psychologist. Like when you go to school and you sign up for that master's and you go to the signup page, how do they define tech psychology? [00:12:38] S0: Oh, well, maybe I need to really ask AI. No, but I think, I will not do that, but I think it's really to understand the theory Human and tech psychology and also the API, I would say. Where does human beings start? Where does it end? And what does human-machine collaboration mean? Why do robots look like us? You know, how to, what is a good user experience? I mean, that we can now chat with our factory. This is something new. And we all know, like from chat GPT moment on, AI has become more accessible. So make tech more human so humans can use it. And I mean, humans develop technology for humans. And on the way, We lose us. And it's so bad because we do all that for us and then we create something we don't understand anymore. And that should not be the case, I guess. And so it was very interesting to study that, but I was also working a lot besides that. So I was not always in class. [00:14:07] S1: Slacker. If you're working, actually, I shouldn't [00:14:11] S0: do that. [00:14:12] S1: Ambitious. We'll call it ambitious.
00:14:14 · Lara Sophie Bothur
[00:14:14] S0: Yeah. And I did both full-time, so I studied in the night and I was not always awake, but just kidding. What you also said before is this wisdom. I don't know if I understand what that means, but I think what's interesting is that With technology, we start to question more and more what it means to be human. And isn't that weird that we create something so far away from being human makes us think more about what we do, who we are. And I mean, we can see that with AI, which we know that once you put in, you get out. But so technology or especially AI is like a mirror. Of everything we are today and everything we have been in the past. And I think today is a good momentum to rethink if we like what we see in that mirror or Maybe not and what we want to change and who we want to be in the future because it's not that the technology is biased, right? We see a biased outcome and we're like, oh, bad technology. But it's just us, you know? It's 70% of the database is biased. We need to rethink if we do everything right before we prompt into AI or before we develop something new, because all that will be our future. And we will always be shocked how deeply human it is. And that has something to do with us. So it's just mirroring us. And I think that's interesting because it makes us question what humanity means and also to define what is the boundary is so weird. Then we say, okay, what does humanity mean? Maybe it's empathy, emotion, trust, And it's like so weird to break it down to those components, because I think, you know, being human is so much more than that, but you need to break it down today to define, you know, Where to embed AI or other technologies and where not to and where we should really not have AI because this is deeply human. But what is deeply human? And so I feel like we're coming back to the roots kind of. Isn't it weird to define what it means and it forces us to do so.
00:16:51 · Evan Helda
[00:16:51] S1: There's so many, so many things. Open the bottle of wine. Yeah, seriously, we should have done this more during a happy hour. It's too early for this level of depth, but no, this is great. There's so many things I want to unpack. So I guess the first thing that's interesting is this idea of, is it good or bad that technology is challenging us and making us ask the question of what it means to be human? Because on one hand, if we're building it right, it wouldn't Beg the question, it would just make it obvious and it would enhance humanity and we would just be on a merry way and everything would be glorious. And on the other hand, thank goodness it's doing that because it's forcing a reckoning. And how do we get on that path? I think the most simple exhibit A would be the Everything that's happening happened with social media and all the cases of Meta and what it's doing children and it's the idea, okay, we didn't build social media for connection, which is a human element. We built it for attention, which is a commercial element. And so, yeah, it could be boiled down to simply that. And then in terms of what you said around Dissecting the human experience and trying to find it. I think all those things you mentioned are definitely the right things. But perhaps the most important thing that's so hard to grasp onto is sort of the things we can't really define. The things that we just feel and intuitively know, but we don't understand. The feeling that there's something more, even if you're not religious. If you're not religious, it's still hard to not see moments of serendipity or manifest something and not be like, that's weird.
00:18:46 · Lara Sophie Bothur, Evan Helda
[00:18:46] S0: But even that we say like, well, that is, um, you know, we, this is something AI can't do because AI can't give things a context, for example, or even intuition. What is intuition? You know, the set of data points, but I think it's so, so bad. [00:19:05] S1: Is intuition a set of data points? [00:19:08] S0: I mean, it can be. It's like, yeah. [00:19:11] S1: I guess it is things that your mind subconsciously is connecting. [00:19:14] S0: It's experience, it's gut feeling, but then gut feeling is a data point that is, how do you define gut feeling? But gut feeling is also a mixture. And yeah, maybe gut feeling and intuition is also very close, but it's also partly it's experience, right? Yeah. I mean, you can't define like feel so hard to grasp. [00:19:41] S1: Intuitions like this is the integration of heart, mind, and body. I think one way I think about it, like the body being the gut, your mind has its own rumblings and you feel something in your heart and soul a little bit about What's right and what's wrong in many ways, or not what's right, what's wrong. It's probably more of a gut thing. I think in the heart, it's more about just some magnetism towards the thing that you can't really explain. [00:20:08] S0: Yeah. There's a physical reaction, same with falling in love. It's like those butterflies you have in your belly. It's like, how can you rebuild that? But I think maybe at some point, And maybe, you know, doctors will rebuild that or like, or they can rest that biologically. Take a [00:20:34] S1: pill for love.
00:20:34 · Evan Helda, Lara Sophie Bothur
[00:20:34] S1: pill for love. [00:20:36] S0: Well, no, but also add that to a humanoid, you know, like that a humanoid can truly feel this feeling, but then like, why should a humanoid feel that? Like who's behind that humanoid is that there's no soul, right? It's like, is that Elon Musk then being in love with everyone? Okay, now it's getting super weird, but yeah. [00:21:02] S1: Well, I think that's where you have to define what a soul is. And I'd love to share with you a couple essays that I wrote because they, interestingly enough, intersect on this conversation right now. First is the AI as a mirror idea. [00:21:19] S0: And the second is