 Alright, so without further ado, Dave, if you maybe want to come up and introduce yourself, this is our entertainment you call, right? Sorry, entertainment that you think? Our moderator, our timekeeper, our whipcracker, he's gonna keep us on the line from this point forward. I step aside and listen to Dave on our time and all those fun things. But the stick, Taylor, because you don't know, and take it away, Dave, and then we'll have the rest of the panel join us. So what are we supposed to do? So we have until 8. Okay. But we only have time for Q&A, so we'll come see how the conversation goes. If we wrap up early in the conversation this good, then we're good. It's whatever we need, but we have until 8. I think that's kind of the final note, too. Okay. Well, thank you, Jennifer. Thank you everyone. As I say, this is super exciting. After how long, how many times are you staring at a camera on your computer and looking at people in little squares who think that because they can't see themselves, no one else can see them. And then they're doing really embarrassing things and you're like, oh my God, you're the one that needs to turn off your camera. Or you hear someone who takes a phone call in the middle of a meeting, right? And it's just like, can you please mute everybody else if you're talking? Can you just do it please? So, I actually am a digital storyteller, so I help companies figure out and tell their story. And that goes all the way from, I do a ton of consumer electronics reviews of work in the consumer electronics space for a really long time. I've been a judge at CDS for probably 15 years. And I have more electronics than you do. Let's just put that out on the table. I have more headphones than you have electronics. Sounds like a challenge. I should have bought some. It's like, grab that gift. I was a mistake on my part. But I also work with companies on things like training videos or product walkthroughs. I'm a professor at DU, so I actually interact with students via Zoom on different time zones, which is a whole other exciting thing. They're all really, really good at not being actually paying attention, as they do in classes too, where they also don't pay attention. But it's a whole other story. For this, we're going to be talking about robotics. And I wanted to start out by observing that today I spent a lot of time dealing with robots. So, first off, I have a robot lawn mower and I left my house and it was mowing my backyard. But I need to do a firmware update. So what did I have to do to do the firmware update? I had to drive my robot to Bertha to the service center because they don't do customer firmware updates. So I had to take it to a tech to make sure you don't rip your expensive lawn mower. So I spent a third of my day babysitting my robot. Meanwhile, while I was doing that drive, I set my vacuum to going. And I came home and it had spent a couple of hours. It has LiDAR. It does 3D mapping. It's very slick, very sophisticated. So all of that's going on the day we're going to talk about robots. And my point with that, and then sort of in a more subtle way, I'm driving my car here and it has automatic lane departure warnings. So as I'm driving, if I'm not paying attention, it'll steer me back into my lane. All of this is part of robotics. All of this is part of the idea that the technology isn't passive devices that we just tell it what to do but that they start to actually sort of participate and take part and improve the process. Now, we have two really great people that are going to be on the panel and we're going to be talking about enterprise robotics. But I just want to start out by saying, I bet you interact with more robots than you realize. And soon enough, you're going to go to the pizzeria and instead of it being a classic Jersey experience where the guy has a cigarette and there's so much ash on the end of it that you're praying. It doesn't fall on your crust while he's working on it. Anyone experience this? That's a very New Jersey pizza experience. It's going to be a robot. You're going to push a button. It's going to be like a DIA. You're going to push a button. You're going to watch it go down to conveyor belt and make a customer. So part of the question with robotics is what is it? What are the implications? What's the ethics of robots? Not just like the three laws of robotics. I'm going to quiz you guys. I hope you know the three laws because you're going to need to. But it's also the question of how much do we want to take away from people? How much do we want to re-engineer and re-think things? I mean, I can be replaced by a robot. I'm sure that the kind of jokes I have are really easy to program. It's just like hashtag dad joke and you're good. But how far do we want to go? So with that, I want to actually invite both of my esteemed panelists to come up. And I'm going to introduce yourself. And I know one of you is going to have to be a DIA. Oh, good job. You volunteered. I knew that it would be killed in fights, honestly. Good man. All right. So Caleb, why don't you introduce yourself first? My name is Caleb Eastman. I'm the CTO and co-founder of Winterwinds Robotics. We have two divisions. One is our consulting division, which does a lot of embedded software development and robotics development for the larger companies, Fortune 500 companies, mostly. And then the other division is our R&D division, where we develop products that we see have market. All of this is sort of designed to culminate in our end game as a company is to develop our own robot for the moon and Mars, particularly for getting water ice off of the poles of the moon in Mars. There's lots of reasons why you want water off Earth's surface. And I can talk about that at any time, but not now. Excellent. Oh, yes, it's Jalali Hartman. So first of all, thanks for having us here. It's kind of surreal to have a live event. So I have a company called Roboto down the street here. We bring technology to life. So if you have a prototype idea or some kind of enterprise robotics system you're trying to build, we can get it off the ground much cheaper or more efficiently. If you go and try to hire a whole team yourself or anything. So we kind of have all the disciplines covered. We have a base platform. We've got to spin these things up quickly. We've created a million mistakes already, so we don't have to. We could make the next. Yeah. So right here in Longmont, eight years. I actually started, got my start going to heat-ups. And I got our first big customer after I moved out of my house. It was five years of my house. Moved to a co-working place down the road within a month. So they walked up to me in our house. I think we met at the first start of the week. Yeah, we probably did. Yeah, I knew nobody in this. I moved here from another state. Everybody, I actually love people here. Great, so let's start out. What's the difference between robots and robotics and enterprise robotics? How do you differentiate those two? Yeah, so, okay. So for me, I'm going to just place a stake in the ground and say that robotics, in order for a robot to be a robot, it must both act upon and sense the physical world. And it must also have software on board. And that cuts out a lot of things off the edges and makes it a lot simpler. Enterprise robotics requires an addition of two things. One is that for me to consider an enterprise robot, it must interact with humans at some point. So the robots in the cage that you see in industrial robotics and robotic builders, I don't consider those enterprise robotics. Because they don't have any, they don't interact with humans. Not the way that you want to. Because you don't really want to go to cage robots. The other thing for enterprise robotics, from my perspective, which knocks out a lot of consumer robots, is that it must be able to interact with and integrate with enterprise systems. So that is an ERPs, SCADA systems, MES systems. Anything that is generally used in enterprise, those sort of integrations, active directories, et cetera, an enterprise robot must be able to do that, or else I wouldn't consider an enterprise robot. And you're differentiating that from industrial and manufacturing. Yes, so I'm actually including a little bit of what people call manufacturing robots in enterprise, in that I mentioned SCADA systems, et cetera. And the integration, if it can interact with people, if it can interact with other systems, I consider an enterprise robot. So I do lead out on the edges here, the robots in cages and the robots that are just designed for interacting with people in a consumer environment. And there's a big hole in the middle, which some people might say there's some of the factories, some of the robots in the back and floor, especially the ones that have collaborative, like UB, et cetera, I would put the Indian enterprise robots, but that's just because I have a really weird view on things. Thanks. All right, too well. Counterpoint. Yeah, so it was like a robotics purist, and he's exactly right about all that stuff. I'm more of a capitalist than a industrial robot. I have the first five years of this business I had a consumer product, we're trying to get a consumer product, honestly, the hardest thing I've ever done in my life. I don't know why exactly, other than you have a lot of competition. As soon as you get something that's working, everybody else can easily copy it. And there's this weird thing where everybody thinks a robot can do all these things that it just can't yet. So they seem like the Jetsons, and they think they're going to buy this robot for like $100, and it's just not there yet. I mean, you guys probably came closest to the mystique that I had. It's like, for example, I had a little Raspberry Pi robot, and then Missy came out. And Spirit came out with a little rover. So that's kind of what happened. You have to be real fast to get out of a lot of money and bring it to market. So when I think about enterprise robotics, I think about enterprise customers. I'm less purist, I'm more, is this device taking some data and making a decision that it wouldn't otherwise? Is it controlling some controls, sensing something, acting, thinking, doing something? But that's where we're at. And we made that shift. And actually what we did do is we said, well, we actually love this consumer product. Let's just give it to people. We put it to schools in Boulder. All the afterschools at the group library, right? We were making money on it. And when I made that shift and said, actually these things that I learned with this little toy, keeping them connected, have them napped, the voice, all of these different things, sending data to the cloud, right? Suddenly I'm like, there's a whole bunch of customers that eat that same thing. And that's where we really, all of a sudden started to make that switch. So I was against it. Like I knew I could do robotics services but I really, really wanted to make this little thing like you see in the movies. And it's hard. Like it's hard. It just is. I don't know. I'm sure some of you are better at it than I would be, but it took a lot out. Took a ton of money. And I think we were kind of laughing earlier. They were laughing at me. It was like, I got into robotics. I started a robotics company because I wanted to start a robotics company. Right? It was a huge mistake. Right? There was no need. I had no idea about any of it. In fact, I went to a new tech meetup early on and presented at CU and people laughed at me. Like I was like presenting this Raspberry Pi and I was like, it's not a Raspberry Pi. It's a Pi. I was all excited about it. It's kind of how I got, but I didn't know what I didn't know. Right? So that's the problem. If you go out and you think, oh, this is going to be great and everyone's going to buy this but you don't know what else there is. And I never once talked to a customer. Everyone sat down and said, oh, this is actually huge. So that's what, that's kind of a hard lesson that I had to learn. Right? But yeah, I don't really, I don't like to find them the same way with Caleb. Caleb, haven't you guys seen the similar stuff that he's doing? I don't think you've shared any of it. I don't know. At some point, yeah, Caleb's one of those like quite, I was actually having a meetup and I was like, nobody showed up and then all of a sudden Caleb's been walking me with this Mars rover thing and I was like, oh, this guy. So, and it's come a long way. I just saw the big one. Cool. The Mars rover, you know, I would, what would you classify that as? Well, so that was, that was, I think you're talking about the proof of concept. Yes. Yeah. So that one, that one, the locomotion is the rocker-voguey-style locomotion. But from a, there's nothing about that robot that can't interact with the people in it and enterprise systems. And so that one can be an enterprise robot. And for me, it's really about how it's used rather than, than the, how it's built. Like so, drones, robot arms, and global robots, ground robots, can all be enterprise robots. Right. So, I mean, I think of the nexus six models from Blade Runner, right, where they were creating to be sort of like ultimate, if you will, robots for extremely expensive environments. So to me, one of the things you're talking about is the idea of having robots be able to sort of augment human experience or human interaction in an environment that otherwise would be essentially impossible or extremely, extremely expensive. Bottom of the ocean at Fukushima reactor, on Mars, on Moon, you know, all of those sort of places where, well, you could do it, but you're going to be glowing orange after a run, and after a month we have to trash it, but that's okay, we'll do it another way. So, I love putting, I love, you know, putting robots in those areas. It's surprising how many situations humans get into because dangerous situations that we get into because robotics aren't there yet, particularly when it comes to it. So, you know, one of the use cases that, you know, we're working on is for a bit of a while then, Firefighter, or know what that's like, and then you understand that it's incredibly very dangerous. It's very dangerous, it's very labor-intensive, it's very, very dangerous, and, and, and so, it's not an ideal place for a human to be, and we're not, we're not really built to handle that constitution. And so, but we also, we also will be there for the foreseeable future. And so, a robot that's operating in that environment must, A, be better than a human at that, at that particular job, and B, we still build interact with humans who are there. And so that's, you know, so, I would actually consider a four, uh, robots that are on the fire line, enterprise robots. Okay, so then, obviously, one of the things that's really important in films is that time delay really gets in the way of you sitting here at Cape Canaveral and controlling something that's that far away. So, is autonomy a necessary ingredient for real lives? Yeah, I mean, they're not really that useful without it, honestly. Um, the, you know, even the Mars rovers are really dumb, like, really dumb. Um, and, and so, the better that autonomous systems, because the nice things about autonomous systems is that they don't lack attention to, you mentioned the Link, the example, they're always one. They can, they never get tired. They never get, um, but, you know, um, there is a whole thing that I could go into around what we call causality that robots are very bad at. Um, robots are very bad at. Robots not understand causality, uh, today. And so, we have to get better at developing, putting, uh, moving causality versus putting probability into robot intelligence. Okay, so, so I'm going to play a, or pick on just a word you just used, and then I'm going to actually sort of you to talk about that. You said understand. So, do you see that in the near-term future we'll be able to understand and make decisions based on the environment and all of that? What do you think? I mean, to me, it seems a little far off. And I think Caleb's probably one of the closest ones. I've not, I haven't talked to him like I said, for me, right? Like the doing, there's a lot of stuff that's like real automated, right? And it's real smart. It all depends on pretty clean data sense, right? And I think what Kim's talking about is up. And it's, these things are hard to even come to because like my whole day literally was going through forum posts, they're wrong, trying to get things to connect, then the drivers, the wrong driver, that's my world. So, when you talk about all these things going to just talk to us and figure out what we need to launch and stuff, I just sound like, okay, maybe one day. But I don't know, Caleb's a lot deeper than that. I'm more of a customer. He's trying to figure out things like, if the thing gets nuked or hit by a balder, it can rebuild itself and figure out where it's at. It's like, well, if it gets nuked, and it's on Mars, then we get a whole other way. There's electricity somewhere and that sort of right. But I mean, I think about like robot backing leakers. How many people have one? Okay, so the first generation of those were awful. And the biggest issue that they had when poop, right? It's really disgusting, but if you think about what a robot backing would do to that, that's 20 times more disgusting. They're still working on that. And so we're talking about these, and I have ones that have LiDAR that can show you maps of how it's like assessed and figured out in my house and at relative elevations and it can sense hardwood versus carpet and everything. And I'm not entirely convinced that if my dog actually had an accident well, I don't know, it looks like sure. So I'm just plowing forward, you know. So you talk about understanding. It's a big duty thing. You can do it. Yeah, yeah, I do. In order to do that, the, I mean, I'm going to make some bold statements here. But, so please feel free, please feel free to take it with a grain of salt. But almost the entire artificial intelligence machine learning community is wrong. Because they're all, there's a huge focus on neural nets and proving neural nets which is fine for classification, which is fine for detection of problems. But that's, they're not, that's not a reasoning. That's not something you can use to reason. So in order to do reasoning, you know, there is the way, you know, the way humans do it. You know, there's three levels that call three wrongs and intelligence. One is associative, which is X, if I see X and Y, and that's, that's a, that's where robots are today and where, you know, and that's, you know, that's where neural nets are today and that's, you know, a lot of the AI community, email community is focused on. Level two is what's called interventional if I do X, I know that Y will occur. This is where talkers are. This is, but the last one, which is where thinking adults are, is what's called counterfactual thinking. Which is that you could imagine what it's like, what the world would be like if you were one foot taller. You could imagine that. That's counterfactual thinking. That's, they're being able to imagine things that aren't here in the world. Well, counterfactual thinking happens to be where a lot of our high level reason occurs. Is that we can, we can, we can take patterns that we've seen in the past. We can create scenarios that have never occurred or that we have never seen. And we can still, we can still operate in that environment. There is a, there is a mathematical underpinning to this, what they call causal inference. Which is, you ever want to look at Judeo-Curl, a book of Y, I strongly encourage you to read that. But if we, if we're going to, if we have to, we have to focus on the math as causality. If we're going to get to a point where robots can reason. And right now, the large part of the data science community is focused on the math of probability. And just building more sophisticated models or algorithms that are based on problems. Again, about probabilistic notions. And not only is it very compute-intensive, but it also will, it also won't allow give robots to believe a reason. So, so it depends, to me, it also depends on how long the data science community focuses on the math of probability. To the extent that they do that is the extent to which we'll never get past this sort of platinum graph right now. Okay, so Robocop, who's seen it? So, remember ED409 in Robocop, right? That was like the ultimate law enforcement robot that killed people. Right? So robots and safety. So, if we have it in a cage, as you say, on an assembly line with a yellow line painted around it saying, for God's sake, don't go in that line. You don't know what's going to happen. But, you know, in a modern factory, the robots are moving around. And in fact, even in an office building, the robots might be delivering mail or dropping off packages on different floors and stuff. A little bit hollywoody. But, as you develop robots, how do you bake in safety to make sure that they're not going to, like, spin out of control and shoot bolts off into the wall? Which would actually make this space look even more cool. But, I don't recommend it. So, I mean, the first thing I want to say is like, all technology fails, but actually that's been my experience with some problems with that or something. So that's a little scary. Personally, I've gone a long ways off before we had to worry about any physical robots that come after us. There's a lot of ways. It's just, for me, it seems far better. Okay, but you say that, but the city of New York, the NYPD, actually had about this robot Jaguar thing. We got sea pictures of that. And they actually, people lost their minds. They were so anxious about this, like, robot creature that could run and get involved in a high-state situation that the city just said, you know what, we're done. We'll get out and we'll never use this again. So I don't know that we're as far as you said yesterday. I mean, I don't know what that was, but let's say the Boston Dynamics spot doesn't have a remote control. It's not actually a robot. You see a lot of these things, but yeah. Actually, I don't know that there's actually a whole community out in Baltimore where actually, I'll use it, where AI, machine learning, we'll put it in by our centers. I have people who are small, I built for myself. I'm training it. I'm trying to figure out, I only run around when I go out with y'all or when I beat my y'all. So we're good. We're really progressive. I don't disagree with that all. It's incredible progress, but there's a big step. I feel like from that to this thing's going to get a notion that it's going to hurt someone in a period of time. Well, I'm not necessarily saying that, though, you know, but hackers are already like friends, and not just hackers, but terrorists are also looking at, like, drones. And that's something I'm starting to see in the police, which is like a completely freaky idea. You know, weaponized drones that they can then remotely control and send into a hotspot. You know, in battle, it's not good for us. There's no saying that hackers are taking that over, and the thing just suddenly decides that it stops humans. Yeah. Yeah, well, that's, like, next year. So, what I see, this is what I see, everybody's thinking about the robots you see on the TV. I know, it's a real use cases where AI is hurting us now. Right? I'm not saying hurting us, but it's being used in ways that we don't even understand. And to me, that's a bigger problem. Like, it's going on now. Let's worry about that. Halo's devices they're never going to rear up in my lifetime and come after me. Right? Like, it's just now that you've suggested it. On a phone! On a phone! I mean, it's just, I mean, I mean, at the Jolli line. There's a long, there's this, it's called the chasm. It's just, it's not there yet. I don't know. Maybe some of you guys are better than I am. So, I don't know. I don't see what you're taking on this. So, when we talk about safety, the most, the most likely way that for a robot it comes unsafe is something going, something going wrong with some, some, like, with bugs, basically. Like, there's 25 million lines of code in Linux kernel. There's tons of ways in which something can go wrong where there was no intent behind it. It just goes wrong. And for these, so, luckily, we have been, we have been running, you know, we call control systems as a, a species for now, 30, 30 years, computer control, ways that we interact with our physical world. So, with electrical grid runs on, with water systems run on, and we have all these, we have these really, really involved very painful standards. INC-662443, INC-61580, there's ACIL-26262, there's a whole bunch of standards that are like, I'm going to wrap you for a second, because you two guys back there, we're going to have a quiz later so remember all those levels. Yeah. You can use them. Yeah. We have, we have a concept called safety integrity levels. Uh huh. Where we, we measure the likelihood that a malfunction would cause harm to humans. And we measure the likelihood that there will be a certain number of humans around the thing. And we'll literally say, this thing has to meet safety integrity level four. And this is what they're doing in, this is the ACIL, ACIL is automotive safety integrity levels, which is what they require everything from the software for autonomous vehicles. This is why autonomous vehicles, uh, you know, they're down on a different operating system, like the export for the real, you know, for the real time operating system. Um, there's latency considerations. They don't even use this in computer, the same type of computer, the same operating system that, that those people have ever even heard of. For all this, for all this, because when you deal with this kind of robot, um, and you're dealing with that kind of, that kind of force that can generate, they really can, they're really, you know, we're really concerned about that. So, I spent a huge amount of time just meeting safety integrity levels and latency requirements and reducing, uh, mitigating risk. Just, it's, it's like, it's literally like, you know, 10 times safer than a human operation. Yeah, which is, yeah, that's where like Tesla's right. That's what Tesla's right. Because Tesla's cars are better than human drivers. And so, for what, like a billion miles on the road, they'll have three accidents and two people will die. And people are like, oh my God, we're so not ready. And it's just like, compare that to any human driver. And you'll find that any human driver, if they drove a billion miles, would have killed like 30 people. Easy. Or more. Yeah, or, or more. I mean, that's just my daughter. Sorry, you're not supposed to know this. I mean, I hope to, I totally get it. Way more train miles. This is the problem. You take away that training model, and it's dumb as to be, right? And that's the problem with all this stuff. We're all designing, yes, we can get machine learning, traffic and all that. Yes, we can get this working. But pulling it all together to where it's like, thinking in real time. I don't know. I don't know. Yeah. I mean, it's interesting because Google just a couple of days ago announced the next generation of Google Translate. And they were making a big deal of the fact that when they went to translate a word, they would also look at the word before and the word after it. And it's like, well, what have you been doing for the last 15 years of your translation software? You know, and their translator is pretty good. You know, we're getting pretty good with translation. And that's just a tiny subset of the complexity of like having an autonomous car swing over here. Someone puts like some four boys in the window and then it drives off and comes to my house and drops it off on my front doorstep. The only thing about this is the text. So that's a good example. So Google, that's a perfect example. They had all of the human language to study. Right? That's how they're able to get it good. But what happens if you don't have that and you can't do that and you don't have that nice data stream of words coming in? It gets real tricky, right? Then you have to like, I think it's sad. I think it's sad. Maybe I'll do this. And I think that's where the perception that even when we get customers come to us, we might want to do like some of those tasks correctly, right? But it's not going to do all of this stuff. And on some cables, they're one of the persons where they're kind of don't agree with me on that, right? They're kind of thinking the next generation. I think even literally, like building the systems for the next. Right. Because, yeah, because all of the, yeah, because the standardized way that data science typically looks today, you know what I'm saying, right? It's never going to get fast. Probabilistic, basically, you know, glorified classifiers. And that's, you know, that's the problem with, that's the problem with having an entire cottage industry build up around what was originally statistics. And, you know, they said, you know, there's a very famous term that they call correlations not causation. But you, you look at data science today, it's almost purely, you know, a child of statistics. So. Yeah. And to me, the best example of that are recommendation engines. Yeah. So I just laugh. I mean, you go look at something like Netflix or Amazon video and they're like, because you watch this movie, there are seven completely unrelated movies that you might like that actually are really horrible. And that seems like a pretty constrained space because they have vast amounts of data to work with. But it's just like Amazon recommending another purchase. There's no way for them to know that someone borrowed my account and bought, you know, makeup or something. And suddenly now I'm getting all these makeup ads. You know, we live in a world where there's like this veneer of sophistication. But underneath it, I'm not sure how sophisticated things really are. So I want to open it up for questions. But before we get there, I do, I warn you guys, three laws of robotic. Isaac has them up. Go. I don't know the laws of robotic. Yeah. Can't, can't hurt. That's the first law. Okay. Anyone know the three laws? All right. Are we going to talk for a few years? Can you why look at him in action? Oh, okay. Can you give him like a beer? That's great. So the first law and this I think is still relevant, which is where Asimov just was just crazy bright guy. A robot may not injure a human being or through an action allow a human being to come to harmony. And then the robot must obey the orders given it by human beings except where such orders would conflict with the first law. And then the third law is a robot must protect its own existence as long as such protection does not conflict with the first or second law. So do you feel like that's baked into modern robots? So we're all doomed. I'm sorry. It depends. It depends. It depends. Yeah. There's there are there are some really, really like I said, there's been a lot a lot of place into what they call sanitary weapons. So it's there. Those are kind of interwoven into those. Yeah. And there's some classic ethics issues here too. So like with Tesla, if a car would take out a half a dozen people on a crosswalk or it could run into a building and just take out its driver which choice should it make. Right. And we are now in a situation and in a world where that's starting to be relevant. I mean, if my robot vacuum cleaner says, I don't want to deal with this room. Okay, whatever. You know, I'll just don't kick it or something. But if my car says, hey, there might be people on that crosswalk. I'm going to do something different that you're not expecting. Then that makes me very anxious. So this is true. So I've been in the time of class he got it on top of every device we had and shut them down. So I didn't teach him that, nothing, he looked it up in the world. So that's more what I think about. Because that kid's a fourth grader. You get something like Halo and get on anything. I'm glad they just compared to the fourth grader. I'll tell you right now, it's not secure. I'll tell you right now, do not piss off controls in years. I'll just say that if someone like, you know how an elevator works, you know how an escalator works, you know how everything runs and it all uses the same system. So if someone understands the systems, don't piss, don't piss controls in years. I just want to say you might have just lost plausible that I ever made. I didn't admit to being a control engineer. All right, so if you have any questions or comments or observations, I know there's necessarily people that are robots, but who knows? Yeah. So I had the autonomy versus enterprise for a firefighting robot, a former wildland firefighter. I don't think anybody should be doing that, John, considering the technology that we have. A lot of it is on a really uneven ground rule. One that's where they just let it burn. People can accessible. So when we have a robot for back country wildland firefighters, is it going to be autonomous? It should be at least semi-autonomous. It really depends. So for example, when they're doing indirect, doing indirect job, it's to dig a trench, then autonomy, it can be as many as we have in our house. Right? If it's doing direct fire attack, then it's got to be able to do quite a bit. We say autonomous, it's a big gradient. So the answer I'll give you is that it should be autonomous enough to be able to do a job, to do a specific task that we know that because we know what we want a robot to do in wildland. So they should be able to do that task in a sense. But if it's really going to decide on its own when to do direct fire attack or when to switch to direct fire attack, that I don't think that it I don't think it I don't I don't think that could be considered in the wrong of its honoring. I don't think that robot's right in that scenario is interesting anyway because one of the other things that they're looking at is having drones that are doing like infrared to have consequences. What's the first thing that emergency services have to say? All of you with personal drones don't fly them in this damn area because you're getting in the way of what we're trying to accomplish. Right. And so you know it's like so now we have drones that are supposed to have big did no fly zones. But those are fixed based on like distance from an airport. But if we have a fire and you put it at that level of risk then what happens when that drone gets banked by some individual drone who's taking live streaming footage for the Instagram chat? We call it Roboto Hawks. Right. And so now and so now it breaks its app right because now we're talking let's have drones that can take out other drones which I absolutely admit sounds awesome but unintended consequences so what are they going to use? IFF transponders like they do in airplanes. So you've gone down the actor road here. I wasn't going there. I just wanted to That's great. At least there's like a very clear a couple of clear examples right here there's this adoption curve in robotics where so Caleb was devised this autonomous thing on Mars on the long spot. He takes someone like right. It's like it's not exactly what we saw in the jet scenes but it's actually what we need. Okay tell us a tiny bit about it. Go ahead. Give us a 60 second pitch go. What is it? So single stream recycling you know you throw everything into a single or cycled in it has to go to a big facility and get sorted out into the constituent materials. A lot of people do that by hand so they swirl bots that and pick it out sorted out on material. Because I've always been really suspicious when I go to places and they say we have a universal recycled bin that really is a tunnel that goes straight to their dumpster. Yeah it does. So yeah the big rule is don't put your diaper in recycling because that's one of the things you pick out the most right? Yeah The diapers are really bad for robots. Yeah These groups are really bad What about the big and the dog poop I know it's even coming full circle You see I was taking a job that it's human someone that saved money and they figured out a way to do that really Right Okay but humans don't want to do but there are humans doing this job Right So your robot is arguably taking away human jobs right So four assembly plant 40 years ago versus one today they're different because of the innovation and because of robotics and automation So at what point do you say we shouldn't develop this robot because this is going to put too many people out of work and a corollary of that is that the workforce needs to continually be trained which we in this country do a terrible job So you know I'll use the example there's not it there's literally not enough single stream recycling happening at all because there's not enough there's not enough the ability to remove the stuff that you don't want in a stream or that the stream can't handle the trash that people wish cycling stuff into the recycling like there's just not enough like so you know the way that I single from control systems because people don't think about control systems anymore the people don't even know that the the amount of corons being injected into the water supply in Longmont is being done by an automated system and it happens too fast too repeatedly too much for a human to do it and so a lot of robot companies are focused on replacing jobs but there's so much that we don't even know that because we can't imagine a human doing that and you know just like you know all the all the you know basically all the systems that we rely on to survive we would always always inspect those systems if we could we might send someone around once a week it is not enough so things we just don't do today that robots could do and should be doing versus just you know using them as a way as a means to replace humans and that's I think it really just depends on you know on how it works so that's where I think that robotics have the most you know have the most use in the year State of Colorado has trained who called workers in robotics if you're in a job that a robot could do you might want to start thinking about something else honestly I'm seeing this stuff especially with this everybody's on like the you know the fermenter quarantine or whatever the restaurants can't find anybody so I just saw a thing on the news this guy's like I had a higher robot he literally got zero applicants to serve so he's got a robot making salads he's got a robot that brings it around to the tables right so in a way this pandemic is celebrating some of this stuff right I don't know why exactly but it's we're been inundated I'm not even busy yeah just everybody now wants automation if they can remove that risk there's a person who wants to do it there's just the other side of that and that is like I'm the just the young working on this how do I you're going to try and they're industrial how do I bring it back to the United States the only way I can make high quality yeah how do I do that and you know this actually there's a lot of there's a concept that's well known in the enterprise industrial space some of you may have never heard of called the great crew change you know you guys don't know about the great crew change it is the fact that there's a huge they call the boomer generation there's a huge percentage of that generation that was such a matter experts on a thing and they were all retired and the next generation the millennials and the generation after Z don't want to do those jobs and don't want to specialize in that thing for 30 years and what that means is right now there's a huge knowledge gap and and and in particular a gap where the way that millennials would rather solve that problem is by automating and being an expert in the automation of things and so where robots can provide a lot of the benefit is where subject matter expertise subject matter expertise and the willingness to be hands on has to has to be replaced by by you know something that can automate the the physical part as well as can be worked on by people whose knowledge and so I honestly worried that we won't replace jobs with robots fast enough for there to be a huge bunch of things that we rely on today to work that flat out won't work and we don't have a plan for that and so people worry about jobs you know in most of the jobs that people are worried about being replaced are miserable jobs but also millennials and people in generations won't do those jobs they won't do and so the only way that we could actually get them done or have the society run as we expected to today is to mitigate the entropy that's been caused and that and honestly we're not we're worried about the wrong things it's not the problem is is that there won't be enough people who know to you know run the mission critical system that we rely on I don't know because that's happening now right here I can't get the people that I need I agree with you it's like I know you it's a problem well no it is a problem I mean it's a huge problem to you know and it's happening everywhere it's happening everywhere sort of on the surface and if anyone wonders why it seems like all of these systems electrical grids less Brazilian all these things all these things are happening if it seems like the infrastructure is crumbling it's because it is right and there's if it seems like there's not an upkeep being placed on all these systems it's because there's not and it's only going to get worse and so I don't think worrying about jobs being taken by role models is the problem we're going to have in five years ten years I used to run a store just down the way here people remember it like electronics and so I have found people who were willing in simmers to A to have old like boxes prepared and it's amazing actually you know I don't do that anymore but there are it wasn't just like old serios and stuff people would bring to me old tabulators I once worked on a scoreboard side you know because there's nobody else around to know what that was and it all just gone and chance you can fix it and find it but a diode is it's a what? a dark job? a dark job you know that electric typewriter I've got an old Smith-Garon electric typewriter that was worth you know we could all be in the deadline you can figure that out I'm retired it's it's actually funny you mentioned that because now we're now we're developing electronics such that that we don't even bother to think about my team but for example we're here the we used to you know we used to you know I would think that any smaller than 0402 on a PCB it's going to be hard to hard to work on later so now we don't even make electronics in such a way to maintain we expect to it works it's the component level that I need a microscope yeah it's hard and I have I need some microscope to do this yeah Jennifer yeah well you know you talk about the deficit of the knowledge space and how so what age does that start and there's obviously a huge boom in young generations interested in robotics and what what age does that transition to different schools that make it happen and how do you see is there a way to feel like that and what age is that I mean yes like we've heard permanent interns now from St. Green Golden they're great I've seen it there's a guy here who builds robotics for schools I can't even name I don't know a year that we went over and he he actually said this part of the answer is kindergarten I mean he makes these little robot things teach him how to teach math decision making stuff that you need to know you want a robot but he starts at kindergarten yeah there definitely is no to young age at this point the only thing I would say is that we need as a community to do a better job at teaching some of the other skills I get these very bright I'm sure you have work but they can't sell out their time sheet or like you know it's like they can't do anything else and I've had enough of that like we need to there's three what's that there's three a little too much stuff sometimes but it's only one meeting a week it's not even like just come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come come So as we get to a more sophisticated society, that target changes too. And so I think we need to recognize that we're not going to find that like super geeky 1950s rocket scientists. They don't exist anymore because they don't need to exist, because that world has changed. So have you seen enough of the population coming out of high school and college and technical skills that you're running around in this case? I'll tell you right now, I'll tell you about that. We are just, we're just absolutely failing the education system as it relates to. Because we're trying so hard to abstract things for them, where they're graduating with like Python, you know, whatever. But that's, like, a lot of people just assume that innovation always moves forward. But we know from evidence that the innovation moves backwards, because what happens is we abstract things to a point where people don't even understand how they can work anymore. And at that point, like, when we teach kids, you know, when we teach kids like, you know, higher level programming languages or no code and just logic. And then they, like, I've noticed that, you know, there doesn't seem to be a strong curriculum out there for Ross. But if you go out there and you have tried to get a job in robotics, the first thing that's going to be on the thing is Ross. So how many kids are being taught Ross versus how many jobs out there are there where Ross is the central component? I dance hiring for Ross. And he's looking for Ross. I'm looking for Ross. We're all looking for Ross. But who knows Ross? No one does, because they're not teaching kids Ross. Not even in the first robotics challenge or anything like that. They're not teaching it. And that was what Misty was trying to do, was, like, make it to where web developers, JavaScript, could do a little a lot, because no one's teaching Ross. You guys see a boulder and they complain. They got to spend the first two years teaching kids Ross. So it really is just a misalignment in terms of, A, we got to be careful of abstracting things too much. B, we have to teach people the skills that they're going to need to know to even consider getting a job in the field. I've seen that at mentorship a couple months ago. I think it's a two-part question with misalignment of his words. So on this side, I tell kids, hey, if you wanted to be a football player and you're a little guy, a little fast, 120 pounder, but you're really fast. But the only person you know that plays football, or the only person you know in robotics happens to be alignment. And you go out and try to be alignment, you're going to do awful, right? You're never going to be alignment. So on this side, there's a book, like, it's called Strength Finders 2.0 that helps people get in the right lane. So that helps. And on this side, we have so much diversification. If you were to get everybody here in back of car to say, get the five skills I need for this kid so I can get my job, every hand would be different, which is great because we want diversification with different options. That's how we're going to solve problems. We're not all thinking alike, but we have one rocket chip that went up. It's easier to find rocket scientists because there's only one ship that could all study the same plans. Now we have so much variety. And then there's moving so fast that when this guy was in school, 20 years ago, and he asked those questions 20 years later, all the languages are different. We're not all learning five years of Pascal anymore, at least we're learning languages at school then. Neat. Maybe not the exact one, but Pascal was an academic plan, but we learned logic. And that, like, the kindergarten or the and loops and conjunction, junction, what's your function kind of stuff. So I think it's combining those two and the third one is what jobs are needed. She's an online statistics.gov and the government wants you to make money, so they can tax you, right? It's in their best interest for you to find a job. So if you go on there, they'll give you the 10-year projection of all jobs. And they'll say, this job has a greater chance of growth in the next 10 years. This one is being automated, so I wouldn't do that. I used to be a medical advocate. I got into computers, but I was a low man in the airport before I had to unload on the computer. This is how I was unloading them. I had to spend a month in the figurine. And we just saw those, that automation just wiped out jobs four days closed on medical centers. Medical advocate schools, we call right off the top five here. They closed in the airport, and they just, they had a machine that could do more than 300 people. So if that came right, you can see technology just hit the job after they're fired. I had a question that followed that, and that is, what jobs are you not going to be automated? Which jobs do we want then? They're jobs. There you go, they're jobs. What jobs will be automated? I can't think of one. Why? Well, but there's a difference between automate and automate, well. That's necessarily automated, because automation is different from robotics. So what would we not robot? What would we not robot? It's up to your point. Yeah. It's a requirement of creativity, hard ones that are going to be the last to be automated. Because even today's stuff can't do that stuff. And engineering is creative. I can say that creative activity. The space is so large that you can automate corners of it. And then you can pull up a chip set and go beyond what I had set. And you press three buttons, and it's like, here's your phone, and that one's done. But if you want something truly unique, you're back to being a good old engineer. And that takes some time. So that's it. That probably is the key to the professional services slash systems integrators. Because it will be hard to automate the, I don't know what the, like whatever the problem is, whatever it is, whatever the system you got, or whatever the legacy, whatever, making it all work, making it all work together. It's going to be a while before the systems can organize themselves so well that you don't need people to make the systems organize well. But I'm trying to think, I'm trying to think of other jobs that will, I think that there are some jobs that won't go away for a long time. I would so vote for it, but a lot over most of the candidates. Those should be the things. But like a police officer, we wouldn't, that's not, there's somebody who wouldn't replace him if he were up there. Yeah, I mean I think that, the need for interpersonal becomes something really hard to automate because really robots, AI, all of that tends to be, we've seen this before, so we can do it again. But if you think of like a therapist, ideally a therapist is like, synthesizing vast amounts of data in real time and saying, well, the specifics of what you're telling me I haven't heard before, but here's some ideas, here's some big pattern stuff. And for that to be genuine, I think it's going to be really hard for that to be a little Eliza robot sitting there on the other couch. Anyway, it's a thought he was talking about, but the job that you would want that isn't going to be automated, how do you figure out when that is and when it will be automated? So fixing VCR, probably not. I'll give you a bucket for that. Now for main TV. Yeah, so one last question, let's end it with a great question, a lot of pressure runs through. Last question. I wanted to know what you were saying about Google Translate, which was already good and didn't get any better, but it's also like, you can't plug a 50,000 word novel cut and paste it into Google Translate. You're going to do a halfway decent job, but at some point it'll get all bogged down. I'm wondering if that's kind of what you guys are driving at. There's only a certain level of robotic autonomy that can be achieved right now and is there a parallel between how Google Translate is limited and its efficacy. It's great with single words, single sentences, but you start getting into passages full pages, captain pasting and it starts to screw up. I think it's a great example because that is one of the biggest best tech companies in the world working with the best set of data they possibly have with all the computing resources they have and they still can't do it. It's still not where we want it. So that's an example. Everything else is going to fall long, many years behind that. Again, they have the words. They know what the words are. That's a huge thing. Thanks to you here. I suspect the reasons are a lot more pedantic and that they didn't make it do that. They didn't make it to translate. They didn't assume that people would translate. So because the truth is, the classifier that they're using, that they're basing it off of is really good and could do it. They could do it. But they didn't make it to where they didn't assume it. So whatever falls apart there, it's based off the fact that the product people at Google didn't think that those people would use it that way legitimately. The underlying algorithm, there's no reason why that thing couldn't do a whole book in a second and get it all right if there was a reason to do that. So that's because the algorithm that they're using is open source. You can do it yourself. You could make that product. No, it's there's four competing ones that are basically the classifier trained on some insane amount of text and stuff like that. But it really comes down to there happens to be a lot between that algorithm and you and a lot of decisions they make. I suspect that someone at some point made a decision that they were only willing to try to accurately translate up to a paragraph and after that, you're not trying to get them to translate a whole book. And so that's that's kind of hard. The fact that they're using neural nets, the fact that they're using neural nets as a classifier is fine but that doesn't actually is part of the problem. Those things aren't built to not reasoning algorithms and they're just perception. They're a place of sense basically and so they're good at that. So it's a pre-program so let's do think about it. They're on that. They can't think out there and they can't go out there. It's a problem. It's all it's doing is calculating the probability for that this is what you're trying to say and like you said the problem is is that they're not calculating it out to where they're trying to put it in context up to a couple of sentences. They don't assume that you're going to want a whole book in context because what's the size of the market to do that? Usually if you're trying to go online you're trying to get a couple sentences for whatever reasons that's the reason. Tomorrow we're all going to wake up and do this really happy because I really had a lot of fun. Alright and with that I will say thank you all very much especially thank you Jelali and Caleb for this super interesting conversation come to New Tech again sign up with Jennifer and comment check out this space again I got the tour you should too there are secret crypts in the basement and stuff we can't talk about but there's ghosts there's like names of who knows what's going on here. I told her I said 23 showed I have only 20 left but that's okay we just left a few so yeah Heather we got for a tour if anybody wants to take the tour and you left over so it's okay. There's some containers over here you can take it all and the food doesn't get better now alright thanks everyone