Timothy Busbice builds AI by copying the wiring of simple organisms like fruit flies and worms. He's simulated up to 20,000 neurons and found that fruit fly brains contain navigation cells that could guide robots. The real problem: we don't have enough connectome data, and most AI research ignores touch, smell, and movement in favor of vision alone.
Chapters
Hello. Hello. Boy, man, something that should be so easy got to be so hard. I don't know why.
I tell you, you know, like technology, man, it, uh, you can never underestimate the power of user experience.
How you doing?
I'm good, brother. Just, uh, Just working away, you know, trying to, trying to make a dent in the internet.
Yeah.
Always, always the case. Look at that shirt right there.
Hey, yeah, I wore it special for today.
I love it. I see, uh, some robotics in the background, I think.
Oh yeah.
Over your right shoulder.
That's there, yeah.
Are you, uh, are you home office or is that garage? I can't tell.
Home office.
Got it, got it. What's new? What's the latest, man? Tell me all the, the latest happenings in next-generation artificial intelligence in 5 minutes. Go, no pressure. No, just kidding.
No worries, no worries at all. Yeah, so I mean, the biggest, you know, hurdle I've I think I've overcome finally is, um, the trying to emulate the larger nervous system.
Mm-hmm.
And, um, I did a test over the weekend, what I call functional applications. So what I'm doing is I'm breaking— so before I did individual apps for each neuron, and then I said, well, that's not sustainable. So what I try to do is create a single program where I had an adjacency matrix with the entire connectome. But that gets so huge and so big. In fact, I found out once I get up to about 18,000, 19,000 neurons, it won't work anymore anyhow, based upon memory and everything else on your computer. So I figured, well, let me try one more thing. And that is to break up— to do one— take, take the large matrix and break it up into small groups. Yeah, like on C. elegans, you know, I use that as— because we know how it works, so it's an easy benchmark for us. So like, I took sensors—
300 minus 2, right? Neurons. So 10,000— was 10,000 the fruit fly? Which means you break scale. Like, if you could only do an adjacency matrix of 18,000, that's one tiny organism of complexity up from a roundworm.
Well, the neurons I have for the fruit fly are just over 20,000, and that's just part of it.
It's not all of it. What percentage is that of the fruit fly?
Probably about a fifth or—
20%?
Yeah.
So it's like 100,000 neurons in the fruit fly, we think, in the connectome.
Yeah, and millions of connections.
Yeah.
So anyway, so what we're, you know, like the test I did had one program for sensory neurons, one for, um, the interneurons or kind of the proto-brain or cortex, and the other one for motor neurons. Yeah, it's working out really well, and I can even break that further, uh, if I wanted to. Because we could do left, right, or, you know, dorsal, dorsal, you know, those kinds of things. But, um, now I, I feel a little bit more confident about the fruit fly. So what I've been working with the fruit fly, the Drosophila, is the— what's called the central complex. So there's, there's 4 things that allow us to navigate the world. And humans are the same as Drosophila. There's place cells, grid cells, border cells, and head position cells.
Mm-hmm.
They all work together like in the hippocampus of the brain to kind of give us the ability to be aware of our surroundings and what we're getting into as we get into it. So I can now extrapolate the neurons for that and then apply that to see how we're gonna get further to neurobotics.
So like, where's— where do you feel like is the biggest blocker right now for you? Is it the—
Not having the network.
Just can't— it's just a scale problem, like the computation?
I think I've got that resolved now after this past weekend. The scaling is doable. I mean, there are some things you have to keep in mind as far as— I mean, obviously, if you put a functional group onto a super fast computer and another functional group on a super slow computer, Yeah. you're going to run into possible issues. But so far it's looking really good, looking much better than any other way I could do it. And then the biggest problem I have now is just the data, just not be able to get data. We're still just— our C. elegans is still the only fully mapped out connectome that exists.
Yeah, so it's like the lab. The labs have yet to map full connectomic sequence, for lack of a better word, for higher-order organisms, more complex than a 302-neuron worm, earthworm. Like, that's nuts. And then what, the whole like Genome Project like back in the '90s? Now they're sequencing your genome and some spit for a couple hundred bucks.
Yeah.
Supposedly, you know. So it's like, I just feel like there's not enough There's just not enough people like working on it or enough funding?
That's the interesting thing. There's a lot of people now. There's a whole, um, kind of— I mean, it's always been there, but it's become a field unto itself of, um, neuron networking.
Like computational neuroscience?
No, it's actually just mapping out the neurons, you know, just exactly what I need to continue on. Um, and then, you know, analyzing those networks within the nervous system. So that's really caught on in the last couple years.
Yeah.
And it's starting to really measure up. And I have to believe there's probably labs out there that have a lot of this information, they just haven't shared it online. So that's, that's the problem. If you can't go to conferences and things like that, you— it's kind of difficult to make You know, I've written people emails, never hear back from anybody. It gets kind of frustrating after a while. You send, you know, ask people to help you out and they just totally ignore you. So I don't know if it's because they don't— I'm asking for more than what they can provide, or you would hope they at least write you back and say, you know, thanks for contacting us, but sorry, can't help you out.
Yeah, like the scientific community is traditionally viewed as supportive and open and sharing research. publishing research, and it's like, here's a group actually trying to build something using the information, you know? And, uh, yeah, it's like, yeah, it's a weird, it's a weird, weird thing. It feels like, yeah, it just feels like you need a bunch of VC money, go grab your own lab, and a bunch of like really high-powered researchers and some equipment and start mapping out this stuff. And in real time, you've got the lab, which is providing the data, the team mapping that into the artificial connectome that you've been working on, and then another team mapping that into the robotics subsystems. And, you know, you just go vertically integrated all the way through. So. Yeah, but I mean, you've been doing tests based on your own extrapolations of what, like, the datasets that are available. And I also wonder on the computation thing, like, it's always struck me that it just seems so similar. Like, the internet is a connectome, and all this talk of Web3, right, like decentralized computation, Ethereum, Bitcoin. It's similar stuff and it organizes similar. I mean, life is a fractal, right? The universe is a fractal. So I, I just know. I mean, you just get these tingles, right? Like the Spidey sense tingles after it's like you see the same thing in different domains over and over again, and it's like a matter of time. And the fact that you've been working on it for well over a decade. Like, it's just, just a matter of patience and time and it'll all come together.
Because, like, I've thought about doing like a SETI kind of thing, you know. Uh, like I say, you know, you could use the computing power out there. There's a lot of other companies that are doing that kind of stuff. I mean, I mine Bitcoin, you know, just using one of my computers that I happen to have a GPU on it and I don't use very often. Um, but I mean, I make, you know, I'm 10 cents a day or something, or maybe, maybe 50 cents a day or a dollar a day or whatever, but depending on where it's at, right? But, um, uh, you know, it's just, it's a— but again, the problem you run into, and there may be a way around it, I don't know, it's like you got some guy out there running something on an old Raspberry Pi and you got somebody out there running something.
Yeah.
i9 Intel computer, how do you, you know, sync up compute cycles?
Yeah, yeah.
How do you balance that out as far as a nervous system? You're gonna have— you could have a really lopsided nervous system. Yeah, running, you know. So yeah, I've got about 3 or 4 computers I try to run things on all the time, you know, at the same time.
Yeah, and that's what's super interesting that I feel like not I have yet to see a single blog post, article, comment, video about computation speed of decentralized networks at each node and how that varies and how you triangulate it. Because they're still, they're still at like level 1, like the first level of Mario Brothers.
Yeah.
They've yet to go to level 2 or like warp to level 6. And like, like biologic intelligence is like level 6. Like you need the warp, warp drive, right? But You're right. I mean, if you're going to build an internet computer, you have to account for the fact that you're going to have different computation speeds plus latency around the world, right? So I mean, even the whole self-driving car thing, like you're just going to run a million parameter like model. I mean, that works on-prem on the car, I guess. You're still having clocks, like calculation time there, but Anytime you want to call out to the internet, yeah, you're done. You're done. Yeah, yeah, yeah.
No, I— it's funny because, you know, a lot of schemes with self-driving cars have modeled into using the internet over cellular data. And, uh, you know, of course they thought, uh, 5G was going to, uh, make a huge difference, but we haven't really seen anything really happen there. And you're right, I mean, so you're downtown New York And you don't have any cell data network because you're between these huge tall buildings, you know. Yeah, so yeah, I mean, there's that, that, that concept really is, uh, there's still a long ways to go. I mean, I have Starlink up on my property.
Oh, you did get it?
Yeah, it really works good. I mean, isn't, isn't it like from there What's the bandwidth down and up?
Isn't it like 50 megabits or something?
It's slower. Yeah, I forgot. Uh, I know I've done speed tests up there, but I forgot what it was. But it, yeah, it was, it was, you know, double digits, but I don't remember what exactly it was.
Right.
I mean, I, I can do— I, on my phone, I get a live video feed from up there, so it works pretty good.
Do you have— so do you have solar power yet? Do you have solar panels?
And I have a wind generator and solar power. Yeah.
Yeah, at the ranch or whatever. So this is interesting. I mean, you're really living in the future because you've got solar power, Bitcoin mining, connectomics, space internet, robotics, you know. I mean, like, the only other thing you need is, uh, like a water filtration system and a farm, and you're self-sustaining.
Let me tell you what else we do, because I live here in California, right? So we're under extreme drought conditions.
Oh yeah, true.
Yeah, well, I have a dehumidifier.
Oh yeah, to take it out of the air?
Yeah, yeah, I get about 4 gallons every day.
Wow.
Water plants, you know, water plants.
And is that— I mean, that's just straight? I mean, are you collecting any, any dust along with that that you have to filter out?
Well, they tell you, no, just don't drink it straight. Yeah, I mean, the water is just— I mean, even when I, you know, take the water out, it feels so nice. and clean and pure, but I, you know, you don't know what's inside of it. And there are filters. In fact, I just got a message yesterday, you need to clean the filters. And so you can see the filters that, you know, get a lot of dirt and dust. They're like, they're like really tight screens, but they get a lot of dirt and dust, dust on them. Yeah. And I'm thinking about up at the property, put up what they call fog fences.
What are those?
It's like you put up a big What I'm looking at, putting up a big plastic fence. So in the morning when the dew— because it does get really dewy up there, you know, the water does. I mean, you walk up, you can sometimes in the morning go out and the grass is just all wet. So this captures the water and just lets the water drop down into a trough and collect the water that way. They actually use those out in the deserts, out in the Sahara and stuff like that.
Yeah, super interesting. I mean, it's like, it's like obvious. I mean, it's like, duh, it's— we could use our planet sustainably. We don't have to—
Yeah, exactly.
Hurt, hurt the Earth. No more hurting the Earth. Yeah, so I gotta run. I know we had some technical issues getting started. I've gotta run here.
Yeah, no problem.
In the next couple minutes, we'll have to do, uh, like a Part 2. What's your schedule like in like an hour?
Oh, I'm available. Yeah.
Okay. Do you want to just reconnect in like an hour and then we can—
Yeah.
Just keep going a little bit?
Yeah. No, just send me a Zoom link and I'll jump on.
Okay. It'll be— yeah, I guess I'll have to send you a different—
It's only 1:30 here, so—
Okay.
Yeah, I work until 5:00, 5:30, 6:00 every day, so it's I'm just hanging out and doing, doing my stuff.
All right, cool. All right, let me— I'll shoot you a text and, uh, and we can jump back on. I'll just shoot you the, the link. If I shoot you the link through text, will that come to your, your computer to do the call?
I can mail it to my computer and then—
Okay, so just email it to you and then just ping you?
Yeah, yeah, no, email whenever you're ready. Just email it to me and I'll, I'll jump on.
Okay, cool. Sounds good. All right, brother. Yeah, talk to you in a bit.
There we go.
Hello. Round 2 commences. Since last we talk So what, um, what's the latest you got going on the robotics side of the world? Have you been playing with any new toys? I know you, you've bought some over the years.
Yeah, I think I, I counted last count, I had like 15 robots altogether. Um, yeah, I mean The thing is, I've been working so much on the software side that I haven't had much of a chance to play with the robotics.
Yeah.
I, um, but I, I started getting back into it this weekend because I had to test my functional application programming. I had to connect it to a robot to— I mean, I did— you don't have to, but you kind of really see if it works or doesn't work. You want to connect it to something to see if it actually You know, does what you expect it to do.
Right.
So I started playing with some robots again, but yeah, I haven't really— you know, our inside joke is that you spend 90% of your time on robotics and 10% on the application stuff. So robots are such a time suck, it seems like, that I— I mean, just to get started this weekend, I realized, oh crap, I got to charge all the batteries and Mm-hmm. All that kind of fun stuff to even get started. So it, um, it just, you know, like I said, just the time. So I was working with the robot that I had bought, and I'm not sure, I think the company might have gone out of business or kind of fledging right now, but I finally was able— it took me a few hours, but I was finally able to get access to it. But then I realized it doesn't really have the sensory capability that I need to make it work the way I really need it to work. I mean, it's got a camera on it and that's about it.
Yeah.
One of the, one of the major, major problems we find with the type of AI that we're trying to develop is you have to have a rich sensory environment. And that, it's almost kind of funny to me, I get into a lot of discussions about this online, on Twitter especially. And it seems like people— well, first of all, people want to basically just use cameras, a camera, and that's it for the whole sensory experience, even on really sophisticated systems that they're trying to develop. And it's like, you know, you just— I mean, how do you expect to get to AGI If all you have is a camera, it's like, you know, take, take, just have your brain and your eyeballs.
Yeah, no.
To reach, you know, general intelligence by just having eyeballs and a brain, it's like, no, you can't, you can't. You got to have other, other things. I mean, you got to have, you know, pressure sensors and temperature sensors. So that's, you know, I've been trying to create or trying to move more in a direction of a sensory-rich environment, and it's a challenge. Yeah.
What, uh, what are you using? I mean, what do you think is the best approach? Sonar, radar? What other types of sensors? I mean, like, yeah, you've got, I don't know, I guess temperature, pressure, vibration, of course. Yeah, it's like the human body. We forget how much we can detect natively, right?
Smell, sound.
Yeah, like pressure, temperature, moisture. I mean, we feel it. That's— the feeling is the combination of all the senses that you're experiencing. It's like took millions of years for us to come up with a single word that combines tens of billions of data points into one thing. Like, oh, it's humid out. What, what does that mean? Like, break that down into a, into a sensor in robotics. And yeah, oh, it's slippery.
Yeah, exactly. And, you know, people don't realize that, you know, it's funny how the AI community doesn't even— I mean, it's not even a consideration. You know, I've had people approach me. I had one guy specifically approach me about, um, he was creating a virtual environment and wanted to apply, you know, our AGI to that. And I said, well, the problem you have is that All you're doing is seeing the virtual environment from a camera viewpoint and maybe from your character, but that's it. There's nothing else in the environment that tells me what's going on. And without those other senses or ability to perceive those senses, then it's kind of very difficult for us to apply our technology to your environment.
Yeah, the real world is full of— I mean, you can build a physics engine, but can you build automatons that represent every piece of dust, wind, earth vibration, motion of the planets, gravity waves, motion of gravity, dark matter, dark energy? Like, it's a lot to program into a single room. Yeah, I think, uh, I think this whole, like, AR thing, you put digital objects into the real world. I think that will be a big unlock because it forces it because you're in the real world and you're trying to have digital things manipulate it. I mean, LiDAR is a good first step to try to map the, what, the topology of a world, but that's not really all of it. I mean, how many different dimensions Would you— like, if you were gonna build Tron, you know, like inside Tron, like the latest movie from a decade ago, Tron Legacy, that Disney put out, how many dimensions would it take for you to actually build that in such a way that it felt like the real world? The uncanny valley, like photogrammetry, but in terms of every single universal element. that humans experience.
Exactly.
A lot. It's a lot to program. It's a lot of manual time. Need some AI for that.
Yeah. And, you know, it's just, it just, yeah, again, it just kind of blows me away. It's just like with conversational AI, you know, they're just basically just a bag of words. I mean, you're just taking, typing something in and you're trying to analyze what they're saying. And, and, you know, one of the things when I did that AI, um, demo, you know, a few years ago up in Sunnyvale, you know, the, the guys from State Farm were there. And, you know, he said that one of the major problems they have in their IVR, you know, conversational AI stuff, is that there's no content. And that's something that's been on my mind a lot lately, is how, how do we develop that contextual, you know, combination in our core, in our AI to understand the world. Yeah.
And as you say that, of course, I ripped the power cord out of my computer. So now I'm using all of my physical senses. I'm literally using my foot, the toes on my foot to grab this little MacBook, you know, magnetic cord, feed it up through the back of the desk, use my other hand, like crook my neck. Like, yeah. I mean, I guess that's, I mean, that's biologic intelligence, right? Like, can your AI do that? So here's, here's my question though. Why do you think, why do you think that is? I mean, it's been the battle that's been raging forever. Like deep learning solves it all. That's all you need. And meanwhile, like, the self-driving cars still aren't at level 5, but they were predicted to be done. Elon predicted it 5 years ago, and he's the man, right? So I mean, we're, we're landing rockets on barges in the middle of a choppy ocean. So why do you think, why do you think there's a block? Why do you think that there is a block? For every, almost everybody, 99% of the population in technology, where they think deep learning is the endgame. Well, we know there's not because like Yann LeCun has said we need another approach. The best people in the world have said we need another approach, but like then they just stop. Like, yeah, why do you think that is?
What is that situation here? they try to beat more into deep learning. You know, it's like the— you know, it's like OpenAI. You know, they think you make, make a bigger deep learning box and you got— eventually AGI is going to happen. It's just, it's just silly. And but the approach is all wrong, you know, with deep learning. And it's, it, it's still a sensual approach. It's still pretty much a visual sensual approach. As you said, it's like eyeballs in a brain. There's no— you know, I've had this argument going on for years, a lot. And in fact, there's a lot— there's a lot more, um, papers and stuff coming out about it, um, that I, I always flag on my Twitter, that, you know, movement is the key to, to AGI.
Yeah.
I mean, this, this guy— and I, I kill myself for not Not, uh, bookmarking this guy wrote an article and said, if, if you want to have general intel— you want to have AGI, you have to start with the stomach. And it's absolutely right, because our motivation to create general intelligence through an evolutionary process started when we were hungry, right?
I mean, Survival. Uh-oh, internet.
You have to, you have to start from that, that viewpoint and then go, and, and we use our senses in order to achieve that motor activity. But, you know, AI's whole approach has been, again, vision and cortical processing, and that's where it stops. But it's like, no, you got to have a purpose on the other side to what you're doing other than just identifying a stupid image or, you know, a cat in, you know, in a video or whatever. It's got to be a lot more than that.
You know what I think it is? I was listening to you talk and a thought occurred to me. Um, we'll go AI, Bitcoin, and then all the way to, um, investing principles.
Yeah.
AI, deep learning didn't take off until they proved it with the little number identification, right? Until like finally computing hit a good approach. I mean, they were working on it in, well, like the '80s and the '90s, and it wasn't until 2016 about where deep learning— I mean, they even called it the winter, right? It was like AI winters.
Yeah.
Same thing with like crypto winter. Yeah. And it wasn't until they got something that was like, oh, holy moly, that like everyone piled in. The lemming behavior of humanity occurred, right? The same thing happened with like Bitcoin and Ethereum. Like number go up and then everybody piles in.
Right.
And so now you see the same thing with like DALL-E 2 and all of this with like type in a descriptive sentence and it will publish back like a really impressive photo, which is very artistic. I mean, it's pretty cool. But here's the thing about investing. I mean, if you read all of the like best investor texts, it's like basically don't be a lemming. Be, you have to be you have to pick something that not everybody else agrees with and be right. So it's, you don't pick Bitcoin when it hit 50 grand, you picked it at 10, right? And like you picked AI in 2015, 2016, not like today. And so I think it's the same thing. I think it's gonna take, like you said, the labs, basically the data, like We need to map some more connectomes of higher-order organisms. We need to finish putting that in software, which you've got, I don't know, decade head start. We need to put that in robotics. And then we need to have that robotic, a 3-dimensional object operate in a 3-dimensional environment. And just for lack of something better, kick it out the door and let it fail and adjust.
Right.
In real time to, you know, the environment. And that's, That's what it takes. I mean, for me, the big use case, I, I mean, I experience this every day and I don't even have kids. It's like you wake up in the morning, you take care of the, your home, you take care of your house, you take care of the dogs, the dishes, you get ready for work or school or whatever. You do the same thing at night. You're, you know, deal with laundry. I mean, you literally are doing all of these activities that, quote, an appliance should do. You're moving between the dishwasher, the oven, the washing machine, like the shower, the bathtub, right? Like the fridge. These are all like single-use appliances. And all we're really talking about here is a multi-use appliance for everyday life in your home. You do that, you save 2 to 4 hours a day times 8 billion people. There's no bigger market size.
Right.
What do we know? Humans want to save time. Humans are lazy. So it's like, I don't know, like spend $100 billion on yet another SaaS app and, you know, make what, 1% every year? Like, it just seems like a really bad idea to me. You've got, you know, I mean, every signal is pointed in this direction, like AR, VR, AI, blockchain, self-driving autonomous vehicles, autonomous spaceships. Like, you're just missing the one thing that brings all of it together. That's my thought, but nobody will pile in until everybody piles in. So there you go.
Yeah, no, I, I have a lot of discussions. I have, I, I've connected with a lot of other AGI researchers, and, you know, everybody's in the same boat as we are, just, um, doing it on their own and trying to figure out, you know, what, what works. There's a whole side of it that, that is trying to use cognitive AI basically to, to try to develop AGI. But I, I keep looking at what they're continuing to do, and it's not much more than deep learning in my opinion. But, um, yeah, you're right. And, and, you know, the, the classic example I give for what's the distinction between AI and AGI is the tea-making problem. Some people call it, call it the coffee-making problem, right?
Right.
So I could have a robot, I could create a robot to make me tea in my kitchen in my house, no problem, right? But I take that same robot to your house and say make tea, it would have no clue. You know, your stove maybe is going to be different, your, your pots, your, the type of tea and where you store all this stuff and the whole, you know, the whole gambit of what, what's involved is totally different in your kitchen than it would be in my kitchen.
Yeah.
And so that really— the AI would allow me to build that robot or appliance to develop, to create tea in my kitchen, no problem, uh, and I could do that tomorrow. But AGI will allow me to take that same robot to anybody's kitchen and can make tea.
Yeah, you know what's funny? That reminded me, I watched this podcast. It was Joe Rogan and Marc Andreessen that came out recently, like, I don't know, a couple weeks, a month ago. Did you end up seeing it?
I didn't see that one.
It was, it was really fascinating. I mean, Marc's a really smart dude. He's from the Midwest. Obviously he's cut his teeth on the internet. He knows a thing or two about a thing or two. And Joe Rogan's talked to like everybody on the planet with like weird crazy ideas. So Joe Rogan was asking him like, you know, are you afraid of AI? And they had the whole like debate, right? All this stuff. But one thing that came up was Marc Andreessen, like the leader of a16z, like one of the number one venture capital firms, most well respected in the world. He's like, oh yeah, we don't have a laundry making robot yet or a laundry folding robot yet. And then Joe Rogan's assistant who's off camera but always pulls up the stuff on the screen while the show's going. He found this thing on the internet after a quick search, and it was this giant machine. It looked like— yeah, it looked like the computer rooms from like the 1950s, you know?
Yeah, yeah.
And it was, it was like this huge machine, and then it would like fold it. And they were like, look, Mark, we've got it already. And he's like, okay, I guess. But it's not really what we're talking about. We're talking about like a humanoid Robin Williams character sitting there like folding laundry for you in the home, and then he goes to make a pot of coffee, and then he like walks the kid to the school bus, makes lunch, you know, cleans up some dishes, put it in the dishwasher, cleans up the house, take the dog for a walk, whatever. Um, I just thought that was really interesting because it's It's one moving so fast that not everybody can keep up with it, even the people whose entire full-time job and hear everything about everything still doesn't know the bleeding edge, like the stuff you're working on, because how could they, right? I mean, you need to be deep in this world. And then the other thing that got me about it was, um, just the fact that like, yeah, the robot is 1950s computer. The difference is that curve of like, it took 1950 to like, I don't know, 2007 before we changed that into something that was mobile and portable. The difference, that curve was like, that line was very shallow. The difference now is the line's very steep. So it's like the tip's gonna happen much, much faster. And I don't think people are ready for that because—
Yeah, there's some scary aspects of it, that's for sure. I, yeah, I'm sure we're going to have a lot of Luddites out there when the technology does come about. Um, but, you know, that's, that's true of all, all change, I assume. Yeah, I, you know, but I mean—
The other thing too is I subscribe to Google Alerts on connectomics or connectome, I forget what the keyword is, and I get an email once a week with all the new papers that being written. And it's not like it's just one or two. Yeah, I mean, it's so many that I can't even— like, I don't even— I can't do that. I go, like, got so much stuff going on. So there's a ton of activity in and around this space. And—
But nobody applying it to like we're applying it. I mean, right? Yeah. I mean, what they're using it for is for, you know, diagnosis of Alzheimer's and you know, prosthetics and things like that. I mean, it's certainly great research and needs to be done, but no one's like saying, oh, we could put this in a box and make it, you know, intelligent— make an intelligent box from it.
Yeah.
No one has taken that leap. That— I mean, it's very— there's a— there's one other company I found out there that's kind of doing the same thing. I think it's called Optum or Optium out of Britain, out of England somewhere. that is looking at the same kind of concepts of what I'm doing. They're looking at like the insect complex, central complex, and, you know, for autonomous navigation and things like that. But, um, but I, I don't really see they've gotten a lot of traction, to be honest about it.
Yeah.
But again, the other difficult part I find is that it's kind of hard to figure out a use case that Would wow people?
Oh, I already know what it is. I mean, I just described it. It's, I'm gonna save you 4 hours a day every day and you're gonna pay $200 a month. You want it? There's a— it's Thanksgiving with those $200 TVs. Oh yeah, no, yeah, lines snaking around the block. Like the market size on that is everybody on the damn planet. Yeah. So like I already know what that use case is.
Yeah, no, I agree. I mean, if you could have a, you know, home servant, so to speak, it would be, uh, it would be nice.
Yeah, you know, the, um, I just lost my train of thought. Where was I going? Oh, I know what it was. I think the other issue is that it's— I, I think people tend to stay discipline-specific because the experts, the true experts, go very deep, right? It's like this concept of T-shaped people and like call it just tic-tac-toe shape people. All of our innovation in the future is gonna come from us sharing what, sharing everything between disciplines and skills. So you, the issue here is you've got the medical area, right? Like lab technicians, like lasers literally going in, you've got software engineering, you've got robotics at a minimum, those 3 things. But I also see that like bioengineering and biotechnologies and connect and like computational neuroscience. I just see the interest in that by the kids in college now. It's, I just, it's coming, you know, it's like there's this desire to want to connect things that haven't been connected before. So it's only a matter of time because just curiosity is taking us down that road.
Yeah, well, the colleges are catching on. I, I had a conversation with, um, somebody out of my, my alma mater, uh, University of California Riverside, and, uh, a couple years ago, and she said— she told me about this, uh, what they call interdisciplinary something, I don't know.
Um, oh, curriculum? Yeah, yeah, where you do this thing?
Yeah, it's like, uh, they created this center where they— the concept is to pull all these different, you know, physics, chemistry, biology, and everything together to develop, you know, newer technology. But rather than being in your little silo, you're, you know— and that's the thing I run into all the time, and that's the problem I see, is that everybody's in their little silo, right? the neuroscientist is in their silo trying to figure out what's dopamine doing in this neuron, you know. They don't see the big picture. And even the people doing connectomics are doing it for biological reasons, not for a, you know, something greater than that for the most part. Of course, the Computer scientists are all stuck in the deep learning rut. And so it's nice. I mean, I was surprised and happy to hear— in fact, she talked to me about, you know, coming, maybe doing a lecture, but they never contacted me again. Of course, the pandemic hit and that kind of killed most of everything.
Yeah.
I was very happy to hear that, you know, that they were going that direction. that they realize that having all these people in these little silos is not— it benefits tremendously, but it, it could be this much greater benefit if we could pull some of these concepts together. Yeah, other, other things. Yeah.
It feels like this interesting, very long duration pendulum in the human species. It's like our universe is one thing. It's not all these subjects that are separate, like mutually exclusive and collectively exhaustive. No, It's one thing, it's all working together, it's one system. But I think as humans with our feeble minds, we, we've— because it's so complex, we've had to compartmentalize and break these down into different disciplines. So the pendulum has swung all the way to this end and it's coming back the other way and it's at this precipice where we're starting to recombine and connect all these things. And I think the thing, the catalyst for that was the internet. Like we had to connect each other So we could connect ideas and connect these different systems. And, um, yeah, it's really interesting. It would be really interesting if like your research on biologic intelligence helped build biologic intelligence by pulling together all of the disciplines, pulling together the people and disciplines on the internet, you know, like what could come out of that? I mean, that'd be some magical stuff. But that's really cool that, yeah, that the younger generation, the kids in college and even the schools are smart enough to think about cross-collaboration. Like, because here's an example, right? What's business? Okay, great business. You know how to read an income statement, but what's your product? Like, what are you selling? What about humans? Why are they buying? You know, like, you—
we've—
it's like, okay, financial analysis in an Excel spreadsheet, but at the end of the day, it doesn't actually tell you much. It's like you need this other stuff working at the same point. So yeah, I think it's like most things, patience is a virtue, and those who are patient will be rewarded.
Yeah, hopefully. I mean, that's, that's our goal, but Um, you know, and like I said, things are starting to loosen up a little bit. Um, you know, my fear is always that it'll loosen up in the wrong direction, but— What do you mean? Well, it's like I say, I mean, a lot of guys I collaborate with, I guess you could call it that, we have these— we get these discussions you can write a book about, um, on like Twitter and stuff, but it You know, there's a lot of guys working on AGI, but everybody's working on a whole different view, a whole different way.
Yeah.
And, and, uh, you know, I think, of course, and they, they think the same thing, my way is the real way.
Yeah, everyone at the camp is above average, including me.
Yeah, so it's like, uh, you know, it's gonna— it's kind of hard to get people to collaborate. And they just had an AGI conference. I don't know if you're familiar with Ben Goertzel. I think that's how you say his last name. But it was basically 3 days to let him talk is what I got out of it. But, you know, he's, he's created this, this system. But I, I heard him talk about this system 15 years ago. And I really haven't seen a lot more come out of it. But yeah, Gary Marcus on, which really kind of surprised me. Uh, and I've had Gary, uh, I don't know if you're familiar with Gary, but he, he wrote a book about a year or so ago basically lambasting, um, the AI, current AI approach. And he gets in, he gets in fights with Jan and all these guys all the time. And he actually has had a few, you know, debates and stuff with them, um, throughout the last few years. And I met him at Camp Sifu. Um, he was— him and George Church was hanging out. I don't even know who George Church is. He's been on like 60 Minutes and stuff. But, uh—
Gary was the— he's a Department of Psychology at NYU And the founder and CEO of Geometric Intelligence. Yeah, it was acquired by Uber. That's— yeah, that's the— yeah, where we know him from.
But he's been an outspoken critic on AI and in, in general terms. And, um, he's really right now in what they call symbolic AI. Oh, he actually— I take that back— more really pushing for a more hybrid approach. So take, take what we have in AI and add to it this symbolic AI approach, which to me, when I look at what they call neural symbolic AI, it's not much different. I could tell there is a difference, but not a lot of difference from like Watson, which is basically a huge semantic database.
Yeah.
And so And I see a lot of these guys, like what I call falling into that trap where they're looking at semantic concepts to kind of bring intelligence together. But again, what it lacks, it lacks that context. You can't— you can't— I can't talk to a robot and— or computer, computer bot, like a conversational AI or something. And say, oh, the weather's really hot today. And that's never gonna, you know, there's no AI out there right now that could understand what hot is. But every animal on Earth knows what hot is.
Because it doesn't have a sensor. It's basically just doing key-value pairs in a database full of other times other people have said the word hot and then sort of just regurgitates that same thing. But it doesn't It isn't associating the sensor on its body that's re— that's sweating, right, in order to drive the thought it is hot.
Yeah, or feeling uncomfortable because it's hot.
Yeah, you know, here's, here's the, here's the analogy for how AI works right now. Um, I go ask you, hey, hey Tim, is it hot? Instead of me going It's hot because I felt it. It's going to the database where you're the database and it's, it's always asking other people. It never has the answer for itself. That's the difference between biologic intelligence and just regular intelligence. It's like I can come up with my own thoughts based on my own experience. And this other stuff is really just, oh, let me go look at the textbook for this. Yeah, it's basically cheating.
Exactly. Let me look it up in the dictionary. Yeah.
You basically have 2 kids in a classroom. And the one kid's the cheat. He's got like the stuff written on his hand and he's got the stuff under his desk or on the TI-83 calculator. And the other kid's just straight up like, no, I'm just using— like, I know this stuff cold. And that's the difference. And guess which one is faster? The one who has to look everything up or the one who just knows it? The one who knows it's faster. So in my mind, like—
Anybody who's done a college exam open book knows open book is fairly useless.
Yeah.
I mean, unless you know exactly where to go in the book to look at something and you can do it quickly, uh, and typically it's only going to give you a hint. It's not going to explain anything to you, not going to really give you the knowledge and information that you need to, to understand. So you got 20 questions on a, on a test, you know, open book, and you got an hour to do it or something, you know, you If you don't know what the hell you're doing and just using the book, you might get 3 or 4 of those questions answered. But the guy who already understands the concept, they're going to complete the whole, the whole test. And, you know, correct.
Yeah, I like— I'm sure you've seen the movie Ex Machina. Yeah, yeah, yeah, yeah, right. I love that movie. Not the end so much because I think it's— that's just Hollywood being Hollywood.
You need to I agree.
Protagonist and an antagonist, so there's got to be some kind of death fear tension. Yeah, which is why I like Robin Williams' take on the robotics movie. I think that is actually the best vision spec for, for the endgame. But anyways, in, in Ex Machina, like, the kids explaining to the, the robot is the Turing test. He's like, it's the difference between being able to describe everything about a sunrise and actually stepping outside and experiencing what it feels like.
Yeah.
It's the difference between a description and feeling it for yourself. And that's the human— that's humanity, is feeling it for ourselves. It's also the reason we're so screwed up, because we all, we all have opinions and feelings, and now we put them all on the internet. And, uh, we're like, okay, well, maybe we don't like that so much.
Yeah, but that's where I get back to, you know, trying to develop context. And I think that's going to be a real key to the next generation, or AGI in particular, is— and, and I've actually even, uh, been trying to figure out or think of ways to, to make that happen. To be able to record the senses as they're happening. So even like you or I can just, you know, sit at our office or walk around the house or go outside or whatever, be able to record the experience. Be able to record the temperature differences and moisture differences and, you know, sounds and what we're seeing and all those things and be able to record everything at the same time. It's not too much different than what we did with self-driving cars or continue to do. Uh, in fact, for a while I even got on this program where they were looking for people. You could download the app, and I did, and put it on your phone and put the phone up on the dashboard.
Oh, Comma? Was that Comma AI? George Hotz?
I don't think it was. I don't remember it was Comma or not, but there was a company that— it might have been Comma— that You know, you go on a 20-mile trip, you just record the trip as you're doing it, and they use that data, that recording, to, to make their AI smarter for self-driving cars. So they got a real picture, a real, you know, knowledge of how people were driving.
Yeah.
And it wasn't just them you know, driving around. They got, you know, trying to get the whole world to do it at the same time and different places and different environments and things like that so they could train the AI, um, on all these different nuances. And I think, I think it, it— something needs to be done very similar but much, many more sensors, you know.
Yeah.
Record the whole experience at the same time.
Well, it's— yeah, it's like this leap, right? I mean, any new driver, any existing driver, you get behind the wheel, you grab onto the wheel, it starts raining, or you get on a patch of ice, you almost have projected your consciousness into the car because you can feel it when you're pushing on the gas and the, the wheels sort of slip, or the brakes aren't working, or the steering isn't working quite as well.
Right.
You feel it, right? Like you've almost embodied the car as a sensor for your environment because you can feel it through 2 things, the pedals, like a gas and a brake pedal, and then like the stiffness of the steering column, and then how this big thing is moving.
Right.
Which is really fascinating when you think about it. Like, that's insane for us to be able to take on, almost anthropomorphize and control in our mind This like crazy piece of machinery. I mean, it's like what's in Alien with Sigourney Weaver when she gets in that huge thing.
Yeah, yeah, yeah.
That's a great scene in Hollywood. But I think that's the leap, right? Is like it's not just that this sonar sensor is reading like one, two, three, four, five bit, bit, bit, bit, bit, bit, bit, and then you've got like ten thousand of them. It's like translating that stream of 10 different pieces of data into this like human feeling. And is that what you mean when you say context?
Yes.
Yeah, I got it.
Intelligence. Yeah. So imagine just instead of my phone, my smartphone up on the dash, just recording the view as I'm driving, which has value, but like you said, What if I had temperature sensors and vibration sensors and pressure sensors everywhere, um, at the same time, recording all that at the same time? And you'd totally get a whole different perspective on what it— driving, driving a vehicle. And that's why we don't have self-driving cars now. And I'm sure you've seen the press about Tesla. I mean—
Yeah.
There's self-driving— there's actually an ad here in California that they've been playing They call this number to tell, I don't know, tell the DMV or tell the feds to stop Tesla's, um, um, self-driving.
Yeah, Autopilot.
Yeah, actually, this guy created this commercial. He's some, uh, safety engineer. That's what he claims to be. And he says— and they show these, the, the actual footage of— they put like, um, a cardboard cutout of a kid or whatever in the street, and the Tesla just runs right over it.
Jesus, that's not great.
Another— they'll do like a person and a kid and an animal, and Tesla just keeps running over them. Like, guys, this is ridiculous. Why are we letting this happen? And I was really shocked to see them do that, but somebody was out, out to get them.
Yeah, it's so hard. It's like 50/50. On one side, my emotions are like Like, get the pickets and the torches and the pitchforks, you know? Like, and then the other side—
Yeah, yeah.
And the others, because it's like humanity, right? Like, save humanity. The other side of me, the builder side of me, is like, well, yeah, but this stuff takes time. Like, I know we gotta protect our people. I mean, this goes back to my whole mission, which is like build things that help people, and it's It's like, yeah, we're building, we're building, but like, we've got to make sure this stuff is safe so it doesn't hurt people. And I think that's really the, the piece is like, they're running tests in production, but when it fails, it's human life. This isn't like Facebook where if you fail in the test and the app goes down, it's like, okay, you couldn't upload or like your photo or whatever. It's frustrating and I get it. Like we're all on social media all the time. That's painful, but it's not the pain of like hurting a human or an animal.
Yeah.
So I don't know, it's this tough thing as we move into hard tech too and especially like biomedical and like the bioengineering with DNA and it just gets into these very human questions. But I'm just sort of in the mindset like You've got all this money, you've got all this time, like, why wouldn't you start exploring and adding on other safety features that are more adaptive, more real-time, run completely separate and in parallel, and could throw up a signal? It's your balance and check, and you need a different system, not deep learning, not AI, in order to do that. I mean, our federal government, for goodness' sakes, has balances and checks. Like the founding fathers knew, I mean, this is America, but they knew you need that. I mean, I, I don't know. I just feel like if, yeah, zig and zag, too many people are zigging and nobody's zagging. Like one, it opens up opportunity, but two, it makes things less, less risk tolerant and more fragile. It makes it frail. Like this is really what it is. Like your Tesla is frail. Until you get something better and it's hurting people. That's, that's it. That's it.
Yeah, absolutely. You know, it, you know, let me go way out left field and say, what if your car could feel pain, right?
Oh, that's—
or, or understand what pain is.
That's interesting.
That change, you know, again, there's that contextual intelligence, you know. Would it, would it change the way a car would react?
And approach situations. Yeah, I don't think it would just keep smashing down the gas pedal into some physical object because it would, it would associate. I mean, that goes back to your whole thing on like nose touch on the Sea Elegans, right?
Right.
It's like, here's food, go towards it. Here's pain or a wall, you know, right? Don't go towards it. So slow down. I mean, it's just—
Change.
It's a different type of application logic at the logic layer of any app. You've got a user interface, you've got a logic layer, and you've got a database. The end. This is just a different type of logic layer.
Yeah.
But I think that's really smart, Tim. I really do. I think that's really smart. What if your car could feel pain? Would it? Would it? Yeah.
Hmm. Yeah. What if you had an animal brain of sorts involved, and like you just gave the instance that we did with C. elegans, and say, all right, when you get close to anything, you start activating those nose-touch sensory neurons. Yeah, you'd see the car react totally different than just like, oh, sonar, I'm 20 centimeters from object, maybe I should stop, you know. Yeah, it's, uh, Or maybe it's too late to stop.
You know what's interesting? It's like encoding fear and hunger.
Yeah, yeah, exactly. Yeah, yeah.
Yeah. And that's why I think it takes, that's why I think the whole thing around, I don't know, sustainable AI or whatever. I mean, I think it's the people who build it put themselves into the product. So you need to be very careful about who the builders are and are they cut from the right cloth. Because fear could be manipulated. And so—
Yeah, you're right. And we've seen that already with even just AI, is that, you know, they, they put out things to talk to people and stuff like that, and before they knew it, it was saying, you know, Hitler was a great guy and, yeah, kill everybody that's a different color than white. And, you know, it was starting to do all kinds of crazy things because The people that were teaching it were teaching it wrong things. Yeah.
And it didn't, it didn't know, to your point, it didn't have context. Like, it's just, it's just letters. It's just ones and zeros. It doesn't have any, any emotion. It's a weight. Our emotions are weights applied to the ones and zeros. So like, yeah, that Hitler of one, the weight of that is very, very high times a million. Versus the weight of pencil. That's, that's encoded with a 1, you know. 1 times 1 is 1. 1 times a million is a million. Probably gonna want to stay away from that Hitler, that Hitler thing, and talk more about pencils, you know.
Yeah, yeah. I mean, that's somebody, some asshole or idiot out there, you know, putting in the garbage that makes this thing believe, okay, well, somebody said this 50 times, so it must be true. Rather than have any kind of understanding or what, what, what the monster this guy was. So yeah, you're right, fear, pain, all those things have to be contextually built into our future systems in order for us to get really what we want. You're talking about the Android, you know, home robot. You know, if it breaks a dish, it's going to be like, oh, you Who gives a shit in current AI, right? Yeah, but, but if it realizes the value to that and what value means, it's going to do its best not to break that dish.
Yeah, that's Nana's dish. That's a hand-me-down. That's priceless because it stores memories. But this is interesting because what you're talking about is shifting software engineering from programming words, letters, and like routes between web pages, and you're starting to talk about programming emotion.
Yeah.
Which is a weird thing. And like, people who I think don't understand that phrase could become very, very fearful of, wait, you're gonna program my emotions, or you're gonna program emotion into machine? And it's like, well, yeah, but in order to protect us. Like, I'm gonna program the fact that it's not good to hit a human.
Yeah, you know, but, but again, why does the robot— it's just like the dish, right? I can tell the— I can teach the robot the amount of pressure and amount of whatever, the physics of handling a dish, but I can't— I can't— well, current AI wouldn't allow me to explain to that machine, you know, that there's other aspects to that. That if you chip that glass, a human cannot drink out of it anymore because it might cut them.
Yeah.
Or, or that, you know, the plate has value and it's going to cost me $20 to replace it, and then get that $20, I'm going to have to do, you know, this amount of work.
Yeah.
So it has no context again as to why that's important or why that, that needs to be. It just knows, okay, this amount of pressure, this is where I place it, this is, you know, uh, the, the total physics of the whole thing. That the physics is, is in some ways the easy part. And that's what we have today with self-driving cars. We have physics. We don't have anything else built into it. True.
The value of human life, what it means, what it means to Barely nudge a human's hair versus like hit them at 60 miles per hour, right? To the car, there's no— it's like, whatever.
Yeah, it's just programmed. There's an object. Yeah, I hit the object, stop. You know, that's all it's programmed is the physics that, that's involved. It's not programmed to think that there's a human life, that human will grow up to be something or somebody. And, um, you know, there's, there's a tremendous amount of value even beyond the pain aspects of it, of destroying that person. It's what it can mean to the future of all of us.
Yeah, this is interesting. This is like a— you need a universal translator for humans and machines.
Yeah. And yeah, in many ways you're right.
Yeah. And, and honestly, even humans to humans.
Yeah, I mean, that's very true. I mean, there's more of that context, you know, contextual experience. We would probably be a different— you know, it's just like, it's just like, you know, they've done a lot of studies where they've taken people that online were just like beating the crap out of each other, especially politically, right? And then they put them in the same room together face to face and had them start talking to each other. It's a whole different conversation, a whole different—
Yeah.
view. They, they, you know, they may not go away disbelieving everything they believe, but at least they have a whole different view and they have more empathy towards, you know, what's happening or the viewpoints of somebody else. And, uh, again, as long as— what's that?
Isn't that weird?
Yeah.
Like, it's always— I experienced this in 1990. Bore on AOL Instant Messenger in the chat rooms.
Yeah.
And it was so interesting because people— it's almost like, um, there's like this cloth that protects you between you and everything else, like this veneer. And, and it, it almost lets the real kind of be exposed more, which is super weird, but it's such a weird thing. Like, why does that even happen? And then the social— it's just societal, like, social constructs and social cues start to break down. Um, and we're confronted with that now. I mean, that is, that is the whole, like, you know, like, scandal on Facebook and— Oh yeah.
I can tell you those triggers on Facebook, right? And, and there's no, there's no consequence to me saying, I don't know, you know.
Hmm.
You may take it as a laugh, you may take it as, you know, oh, I really hate, you know, very despairing, you know. But I have no clue, and I don't, and I don't care. I don't have to care because— but if I had to feel your emotions and read my response Boy, wouldn't that change the world?
Oh man, yeah. If I felt what you felt when I interacted with you, would we start making each other feel better?
Yeah, yeah, that's, uh, start to think more openly about what we're doing and why we're doing it.
Well, you just, you just figured out the first trillion-dollar business, so congratulations. Well, actually, that's not true. That's Apple and Microsoft and Google figured that one out, and Amazon. This is maybe the, the $10 trillion business, although maybe not. Maybe people are like, yeah, I don't like the way you make me feel, throw this device in the trash.
Yeah, yeah, I mean, I'm sure that would happen.
This is broken, I feel like garbage. Yeah, this thing is broken. It's like, ooh, actually there's not a bug in the product. Um, yeah, fascinating, man.
Yeah, yeah. So I, I— and actually, there's a guy just joined our AGI conference on Twitter. He, he claims— I, I don't know how true it is, but he claims he used— he uses more of a homeostatic, uh, view of, uh, his, his, his brain or his AI. Um, and he, he talks a lot about how he integrates emotions and stuff into his AI. And I'm, I'm skeptical. I'm always skeptical. So just the scientist in me. But, you know, I, I don't know whether or not he has— how valid his product or his development is. So sometimes the things he says, it seems to be kind of really overboard, uh, in some ways. But, um, Um, but I've always said, you know, there's a homeostatic aspect to, uh, AI that should be embedded that we, we, we, you know, we— no one ever considers or thinks about. And but again, and what I mean by that is that self-preservation, you know, I don't want to feel pain, I want to feel good, you know, stuff. I want good stuff to happen and be in my life. I don't want— you know, bad things to happen, pain, painful things to happen. So, uh, it definitely will play a part.
Yeah, that's evolution. That's millions of years of evolution built it into us for survival, and, and we just passed along that genetic sequencing to help us survive. Yeah, it's super.
All right.
Oh man, it's like, it's everything. It's science, art, philosophy, religion. It's all up for grabs.
Absolutely. But it's built into your connectome.
Yeah, it is.
And that's why we're in the right direction.
Well, yeah, I mean, the hard part—
I mean, a lot of all these guys are trying to build something that replaces the connectome, right? And I'm always on them saying, hello, it's, it's already been done. Yeah, just why are we trying to go off and rebuild something that's already been done? We're trying to reinvent the wheel.
Yeah, you know, you know what I think the problem is? We should really look at like Facebook and Instagram because they're great at copying other things. Just you have them start building, you know, AGI because they're great at copying stuff that works.
You know, like it might not be a bad idea.
Like, they know how to copy really, really well. So just give them the tools to copy the, the features that work. Here you go. Here's AGI. Use the connectome. Have you thought of nature? You live in it.
Copy that. Exactly.
Allow me to introduce you to nature. Yeah, please copy away.
Yeah, I mean, the millions of years it's taken evolution to figure it all out, man. It's like You already got a blueprint. Now, we've talked about before, you don't want necessarily everything, uh, you know, the abnormal, you know, scenario, uh, it's always— and that's what you were kind of leading towards earlier, was that, you know, you can— you could create an AGI that no one would want to have any part of, right?
Yeah.
And that— and that's the danger. If we could create an AGI today of intelligence level that, you know, even if it's more like a minion-type intelligence, right? But it could feel, you know, joke and, you know, laugh and that kind of stuff. I mean, you'd give it to some, you know, terrorist group in the Middle East or something, and, you know, that they could pervert the hell out of it to the point where, you know, oh, you're a Christian? Boom.
Well, see, that's why It needs to run on low-power devices on-premise and not be connected to the cloud. Ding, ding, ding, we have ourselves a winner. Yeah. So I mean, that's why the whole like Connectome thing, I mean, what's your brain run on? It's really low power, right? I mean, it's not like we're having, we go to sleep once a day and recharge, we eat food, but it's not like we're pulling in megawatts To power a data center. No, no, this thing.
Yeah, yeah, you know, and, and I— in this functional programming I just completed, uh, I did it in Python with the purpose— or the purpose I have with running it on a Raspberry Pi, or Raspberry Pi, I should get 3 or 4 of them. And so far it looks like it's very doable. Yeah, so that's That swap, you know, size, weight, and power, uh, concept is definitely, I think, doable and in the works.
Yeah, I've seen it. I mean, you had that years ago. Yeah, little RC rover just rolling around the side, right?
Right.
And you had what, it was like a little solar power thing on top, a little mobile battery pack, like a little RC car, like a $10 sonar sensor. I mean, it was all off-the-shelf parts, like, for a reason. And it cost— what was the total dollar cost? It was like less than $1,000, right?
Oh yeah. Oh yeah, easily less. Probably a couple hundred or maybe $300 or $400, something like that.
Yeah, which is totally achievable. I mean, VR glasses, PlayStations, $500.
Yeah.
You know, people have car payments more than, more than a couple hundred bucks. So that's totally doable. Yeah, interesting. Well, maybe that's where we— that's where we leave it for this, this first episode. Because, you know, I gotta take the dogs out because they're biological creatures. I need food because I'm a biological creature.
Exactly.
And, uh, yeah.
And your connectome is allowing you to do all those things.
That's true. I'm gonna stand up and I'm gonna not use any deep learning whatsoever in order to go That's right. Who'd have thought? Who'd have thought? Nature, copy it.
Yeah.
And you don't have to overthink it. That's it.
We already have it. All we gotta do is apply it.
Yeah. I think I'm gonna work— I'm gonna use that as like the title of this podcast episode, and I'm just gonna go cold with it with special guest Timothy Bussbuss.
Oh man, man, love it.
Well, this is great. Like, let's, let's do more of these. We'll do part 2. I'm getting back out there. I'm sick of like not creating stuff and talking to you guys every day. So, um, yeah, let's get the— let's get all the crews back together and talking again. It's way more fun that way.
I agree 100%. I love it.
Okay, nice. All right, dude. Well, love to your wife. Enjoy the evening and, uh, The green sunset in California over the water.
We got a heat wave coming, man. I can already feel it.
God, it's been so hot here in Tennessee, like 100 degrees all summer, and then it finally like monsoon rained yesterday. I walked outside to get my dog from 5 feet from the daycare into the back of the car, drenched.
Wow.
It was, it was crazy. So yeah, I don't know, this weather thing.
I wish we could get some of that here.
But yeah, this weather thing's gnarly. We're gonna need some help from robotics, I'll tell you.
Yeah, yeah, definitely. Yeah.
All right, brother. Well, have a good night. We will talk to you soon.
All right, have fun.
Alrighty, bye.
