Episode Transcript
[00:00:00] Speaker A: Welcome to Infinite Future. I'm Todd Thomas and today we're exploring the innovations shaping tomorrow's world. You're watching NOW Media tv.
Hello and welcome to Infinite Future, where I take the breakthroughs shaping the next decade and make them feel immediate, real and usable. I'm Todd Thomas and this is Infinite Future. Today I'm joined by Alex Roy.
Alex is a general partner of new industry venture capital.
He's also the co host of Autonacast and co founder of the Drive, formerly special operations lead at Argo AI and a man who's lived the difference between theory and reality at a very high speed. And I will say just a little bit more, Alex is legend. I'm so thrilled that you're on. I mean, that's all, I guess, the technical intro, but the reality is it you're an author, you're a podcast host, you're a TV host, you're a car rally driver and the godfather of the modern cannonball.
[00:01:01] Speaker B: Thank you, Todd.
Thank you.
[00:01:05] Speaker A: I'm just, I am so thrilled to have you on. This is a really a pleasure for me. I, I am a fan and I guess I'm a little bit starstruck that you're here in the studio with me. I'm just really glad to have you here. And a special treat for everybody watching.
We're going to have a little conversation here. And in segment two, Alex actually brought his car with him. The, the car that he just recently did, an autonomous New York to LA Cannonball. And he's going to take us outside and give us a tour of the car and show us some of the fantastic tech he's decked out this car with. So make sure to stay tuned for segment two. It should be really interesting what we're
[00:01:46] Speaker B: going to do now.
[00:01:46] Speaker A: We're framing this episode around one question.
How do I know what's real in mobility and autonomy when the incentives reward type.
Your career spans performance, media autonomy and deep tech investing, which makes you a rare translator between worlds.
So, Alex, when people meet you now as a vc, what's the part of your story you don't want them to skip past?
[00:02:16] Speaker B: Well, you know, VCs are generally
[00:02:20] Speaker C: split into two groups.
[00:02:23] Speaker B: The people have actually operated companies, started them themselves, they've experienced scaling and exited, hopefully successfully, and bring those lessons to investing. And then there's the other ones.
And so, you know, I'm a, I'm a fan of, you know, inventors, creators, innovators and builders.
And actually, I'm fine with people who've had catastrophic failures, but you had to have done something, preferably something unique, out of the box and hopefully relevant. But even if it's not relevant to what you're investing in, you at least have like the structure, the paradigm of success and like a rough roadmap of what that should look like and obviously a keen sense of people.
In my case, I put my, I'd put myself in the first camp, but in a very non traditional fashion. And so if people know what I was doing 20 years ago, setting the Cannonball records during a time when no one thought it was possible, if they know, I don't have to explain it.
But if they don't know, then I have to, I unpack that. Our attempt to break the CannonBall record in 17 years ago was literally modeled on launching a startup.
We had, you know, we had, you know, the competitive landscape, which was law enforcement, weather and road conditions. We had, you know, our, the, our burn rate. We had a theoretical revenue model, although that was, you know, book and film and severe insurance and liability issues, including potential prison time. And so when I unpack it that way and I, and then I start breaking down the threat environment and all the competitive threats to success through the lens of startups, it suddenly snaps into focus and makes for a really fun, unique story. Hopefully I don't have to spend the rest of my night talking about cars, which are not that they're inherently not that interesting. But what is interesting is how we got to success. And I always say the miracle isn't that we did it because I was smart, but despite all the mistakes I made.
[00:04:33] Speaker A: Well, you mentioned earlier the founders that have catastrophic failures and I often think the guys that have really successful second or third startups are guys that had disastrous first startups. You can learn so much from your mistakes.
So I guess early on when you were, when nobody believed the Cannonball run was even possible, what were the early mistakes that really you learned from that ultimately made it possible?
[00:05:02] Speaker B: Well, the biggest mistake was making a list of the top 10 things that could go wrong with the car and without looking at what they would cost to mitigate a resolve and then replacing all those parts and taking measures against those top 10 things and then going and breaking down halfway on track to beat the preexisting record by I think it was like three or four hours. We were on course to get cross country in 28 hours, which is an incredible time. The prior record was 32, but we broke down exactly halfway because of a clogged fuel pump filter. It's literally a $10 part now, if I had put costs next to the top 20 things that could go wrong and just knocked out the 15 cheapest things to resolve, that would never have happened. But instead I looked at the likelihood of failure as opposed to the cost of mitigation.
And that was a catastrophically naive lesson, mistake and lesson. And once resolved, we went back out and we were successful.
But overthinking the wrong problems, like paranoia is good, but paranoia needs to be correlated to some risk analysis and cost of mitigation analysis. And I'm guilty. I've made that mistake many times.
[00:06:28] Speaker A: Well, and you're right. As you're telling that story, I can see the direct correlation to any startup with those exact same words, just a different product. Yeah, that's fantastic.
So what did the Cannonball era teach you about systems, risk and constraints that most future thinkers still ignore?
[00:06:49] Speaker B: Well, you know, people believe about things they don't understand, a lot of things which turn out to be completely untrue. For example, in the context of Cannonball records, the belief is that you just get in the car and drive as fast as you can.
Completely untrue. That is table stakes.
But that is not going to get you to success. Otherwise hundreds of people would have done it between 1983 and 2006 when we did it. And dozens tried and they all failed because they failed to use a quantitative, I think, startup lens. And so when we wrote the plan, we went out and tested the fuel transfer rates of pumps at different gas stations. Cause you need to know, you don't know where you're going to refuel, but you know, you have to, you need to have really optimal choices.
You know, we went out and researched, you know, the federal Department of Transportation records on peak travel days and made sure we were traveling on the opposite days. And so that turned out to be really two days a year were optimal in April 1st and October 7th, Columbus Day. And maybe not intuitive, but those are three day weekends which are not traditional high travel weekends. They're outside the summer months. They also happen to be when tornadoes are still happening. And as a result, law enforcement is generally diverted to deal with that. And there are fewer vehicles on there. And so if you start adding up all of these small counterintuitive lessons pulled from raw data and they start compounding as mitigating factors against disaster, such that the driving becomes the only thing you need to worry about. Because you can say, I've made a good faith effort, a best effort to mitigate everything I can research.
And then you could focus on the things that you really can't, which is how fast you drive in this time, in this place. Road surface quality would be another. Tire selection would be another. And so you compound them all. And it's no different from mitigating risk in a startup trying to do something that is allegedly impossible.
[00:09:01] Speaker A: So once you had the plan in place and you learned all of this information and made your choices and you launched and you hit the gas and you took off on that drive, how much were you stressed about your success and how much were you able to actually enjoy the drive?
[00:09:20] Speaker B: It was absolutely not enjoyable.
Didn't even think about it. I mean, thinking about it now, the notion of doing it again in that way is terrifying to me. And I'm 20 years older and I'm a father, so my risk calculus personally is very different.
But once you're in the car, once you're in the zone, it's no different from what many athletes talk about and some founders talk about being in zone, a Zen state where the world no longer exists. All that exists is a set of tasks you've outlined and in the context of all the risk mitigation choices you've made. And so when you're in the car, it's as if nothing exists beyond the vehicle. It feels like being in a submarine, and you are just hoping you remain undetected. And as long as you are, you can execute your tasks to. To the best of your abilities.
And cannonball history is littered with people who thought they had done all those things. They had not come close to doing the thing.
And so I always find it funny when I see a very beautiful hypercar parked in front of a restaurant. And the guy has a.
It's a 300,000, $400,000 car. He's got a $300 radar detector in it. Now, even if he knows how to use it, it's unlikely he will successfully drive that car over the speed limit anywhere without being caught, because he does not understand.
He does not understand what it takes to use that tool to its full capabilities. But he does understand the other half of car ownership, which is, what can I project to the audience of the world who sees me park it in this place, this restaurant? It's a very different task.
[00:11:10] Speaker A: So you're the godfather of the modern cannonball, and you've set car rally records in numerous events.
You've been a pioneer in so many ways.
Now you've built a career. You're really a known and recognized voice in autonomous vehicles. And now you've done something brand new again. And you're once again a leader and a pioneer and you have led the first successful autonomous cannonball across the country.
Phenomenal. And for viewers out there, stay with us. We're going to be back in a couple of minutes and Alex has brought his car with us and he's going to give us a tour and show us some of the amazing technology that he's decked out in this car. I'm Todd Thomas, this is Infinite Future. We have the pleasure of having Alex Royer with us today. And we will be right back. We'll be right back with more conversations at the edge of technology and transformation. Stay tuned. Every week on Infinite Future, we explore breakthrough innovation across every major frontier. We talk to AI architects, biotech pioneers, space entrepreneurs, clean energy disruptors and the thinkers redesigning global systems. We don't just talk trends, we examine scalability, ethics, economic impact and real world implementation.
If you're building, investing in or leading the future, join me on Infinite Future only on NOW Media tv. Because the future isn't predicted, it's engineered. And we're back. I'm Todd Thomas and this is Infinite Future on NOW Media tv. Let's look ahead.
Hello and welcome back to Infinite Future. We are not in any studio. We are in a parking lot on a beautiful sunny day in Arizona. And we have returning with us Alex Roy and he has brought his spectacular car and he's going to give us a tour of this beautiful beast and the amazing technology. So Alex, show us this bad boy.
[00:13:12] Speaker C: So this is a 2024 Tesla Model
[00:13:15] Speaker B: S long range which has been specifically
[00:13:18] Speaker C: prepared to break two Cannonball records and has so far broken only one. And it will remain only one because
[00:13:25] Speaker B: technology has now moved past what this car is capable of doing for the outright electric record.
[00:13:30] Speaker C: But this vehicle was recently used to set the first New York to LA Tesla full self driving Cannonball run record.
[00:13:39] Speaker B: And that means we drove it 100
[00:13:41] Speaker C: miles, we rode in it 100% cross
[00:13:43] Speaker B: country without touching the steering wheel or intervening at all.
[00:13:47] Speaker A: Wow.
[00:13:48] Speaker C: So it has a number of additions which were installed primarily for the outright electric speed record, which all of which proved useful on the FSD record. So on the front of the car we have for safety a Teledyne FLIR thermal camera.
[00:14:07] Speaker B: So this camera is connected to a
[00:14:09] Speaker C: distribution display in the interior. You could drive with no headlights, but you would never do that because it's dangerous and irresponsible. So this is the Teledyne FLIR thermal camera. This camera is then linked to the display on the interior of the car. And this is really perfect for identifying animals and police cars. Highly in the median here you have the sensors for the Stinger system. So this is specifically designed to look like it's factory. So you have a pair of sensors here, one emitter, one receiver, second pair in here on either side of the Tesla T. You can see them hidden in there. And then the third pair is right here.
There is a symmetric pair to all three in the back hidden in the rear bumper. A wonderful thing to have.
It's a great company.
I'm always surprised that a thermal camera is not standard on all cars, but obviously that's a cost benefit analysis. Most people don't want to spend two or three thousand dollars on that option.
[00:15:07] Speaker B: Additionally, we have here, carefully concealed to appear to be a factory option, how
[00:15:16] Speaker C: the Tesla automatic opener works sometimes.
[00:15:19] Speaker B: We have here a pair of sensors from Stinger which is a radar, radar
[00:15:26] Speaker C: detector and laser detector and laser jamming system. So if for any reason and not
[00:15:32] Speaker B: that that would happen, we were struck
[00:15:34] Speaker C: by police laser, this system would jam it. And then of course, it also serves
[00:15:39] Speaker B: as a radar detector.
[00:15:40] Speaker C: All right, so this layout is as clean as I can get it without cutting into the car.
So first we have this is a Valentine radar detector which is, was state of the art, you know, 30 years ago. Updated now, still quite good.
Sensors front and back.
This is the Stinger control unit which is linked to the sensors in the front and back of the car which indicates radar detection, laser detection, and laser jamming. So this is software updatable and the best there is. Frightfully expensive. I'm not endorsing it, but it's very good.
This is the display for the Teledyne Fliro thermal camera.
[00:16:22] Speaker B: This is removable, so it's dangerous in case of an accident.
[00:16:26] Speaker C: So I Velcro it and there's a cable which you connect normally.
This is my favorite, my favorite thing in the car. So this is the highway radar app. And this some crazy cannonball fan coded it. This scrapes WAZE data and creates a heat map of historic WAZE data.
Obviously, it has a database of police speed cameras. Most importantly, it takes the speed data and the heat map and combines it to create an algorithmic risk score for whether or not you can speed at the place and time where you are right now.
Additionally, let's see if we can get
[00:17:00] Speaker B: over to the airport.
[00:17:01] Speaker C: Can we see it?
This will scrape FAA tail numbers that correlate to police aviation and put them on the, on the app as well. I forgot to mention that the best part, it Also includes the location of aircraft whose tail numbers correlate to police aviation. And so that is, I mean, light years ahead of traditional ways. And this highway radar app, combined with a great radar and laser jamming system are an example of a compounding technologies that mitigate your risk of being caught however you're driving.
So this is one instance of ways. This is the one I used to observe. Whether or not this is operating in
[00:17:46] Speaker B: real time, I'm quite pretty sure it doesn't.
[00:17:48] Speaker C: So you'll get alerts for police here before you see them there, but you
[00:17:52] Speaker B: have no historic data.
[00:17:54] Speaker C: This is the second instance of ways I use to actually input police ic because I don't like to do that on my personal account.
You know, if the police use Waze, they know what's going on. And this iPad is used as a backup to everything else in case something breaks or the power goes out, because you never know.
[00:18:11] Speaker B: This is a Nexar AI powered dash
[00:18:14] Speaker C: cam, I guess the best dash cam there is, which does all the traditional dash cam functions, but also allows Nexar to classify road intersections and road surface by quality risk safety, which is, interestingly enough, their product is to process all that data, have you run machine learning algorithms against the video they record, and then resell that data to cities and AV companies to do navigation, routing and road construction planning. And so this is more of something I just do because I love to contribute to platforms that actually make the world, world a better place.
[00:18:55] Speaker A: Fantastic.
All right, so as you're doing this cannonball and you're going from New York to la, you're flying across the country. The goal is to go the entire distance without the autonomous system ever disengaging. You never have to touch the wheel.
What are the biggest challenges that would cause the system to disengage? It would cause you to have to grab the wheel.
[00:19:16] Speaker C: Well, the biggest challenge is the mistake I made once, which is in the course of telling a story and just being in the car, things are going well, just laughing and grabbing the wheel for no reason other than you get excited. And that's happened to us.
But the big one is having sufficient experience with the Tesla FSD version that you're using to know that you can take it on the day and trust it. And as you're going across and encountering scenarios which may seem risky, having the discipline not to preempt the vehicle's decision and interrupt it as it resolves how to drive safely across.
We made four attempts to go cross country with subsequent generations of Tesla FSD software 2 that were an event version 12, one in version 13. And they had, over time, a vast improvement in involuntary disengagements, which is what you want to avoid the system spontaneously deciding, take over. Now that can be really alarming, it can be dangerous. And on the fourth drive, it never happened.
[00:20:35] Speaker B: Wow.
[00:20:38] Speaker C: Caveat. It did happen once, but this is, you know, this is what. When you deploy software like Tesla, full self driving at scale in the real world, without a clear operational design domain, meaning you can use it anywhere, you will encounter things, users will encounter things that aren't in manual.
For example, we were, we learned very quickly because we went across in winter, that the vehicle camera, the rear camera, after two hours of driving will be dirty or iced over. Therefore, the vehicle could not autonomously park itself at a charging stall.
And if we intervene, we have failed.
So we'd have to trick the system.
And you trick it by navigating to a point 50ft from the charger.
It's parked, a guy gets out, you clean the camera, you engage the system again, and the car is still 100% autonomous while in motion.
However, during one of these recharging schemes that was unfolding, the vehicle was backing up and we got the red hands of death, the involuntary disengagement warning. And in every prior incident on prior drives, you're in motion, you're going 50 or 60 miles an hour, 80 miles an hour, you have to take over. This is safety critical. But we were going zero miles an hour. We hit engage, it began to move, then it stopped. Red hands of death. And so we sat there and I'm like, what if there's something we don't know? It's not documented in Tesla software. What if we just let it play out, see what happens? And we waited and we waited. After about 45 seconds, the vehicle resolved the issue and proceeded.
So having the discipline and faith in the machine at the right time is everything.
[00:22:40] Speaker B: Wow.
[00:22:41] Speaker C: No different from trusting a person. There's. Sometimes you can't trust the same person, be extraordinary in one scenario and a disaster in another.
[00:22:53] Speaker A: So it took you four attempts?
[00:22:55] Speaker B: Four attempts.
[00:22:57] Speaker A: And you said there's two optimal days a year. Does that mean it took you two years?
[00:23:01] Speaker C: So to be clear, an internal combustion cannonball. At max speeds, there are two optimal days a year. But for an electric cannonball, you have a range of days. The electric cannonball, optimal speed is in the high 60s or low 70s. And so you're, you're not breaking the speed limit most of the time.
So it becomes optimization of Ambient temperature and to some degree, traffic and availability of superchargers when you arrive, which means you have, you know, probably 60 or 70 days a year. You can do it. In our case, we were pressed for time because we were aware that others were thinking of doing this. And my competitive spirit got the best of me and I'm like, we leave tonight. It doesn't matter what time it is, we go tonight. As one of the largest winter storms in decades was unfolding, and so we opted to navigate through it in between storm fronts, which was.
It could have been very dangerous. We got lucky and were successful.
I must add full credit to David Moss. David Moss successfully drove cross country in his Tesla about two weeks before we did. But he took the southern route, not the cannonball route, 30% shorter. And I think that's great. But the Cannonball route is the Nurburgring of autonomy.
That is the route you have to take to set a benchmark against which any other attempt can be measured.
[00:24:32] Speaker B: So you did it.
[00:24:33] Speaker A: You were the first one. You mentioned in the Tour that this vehicle now has some technology that's been usurped.
Will you get a new car and do it again?
[00:24:43] Speaker C: Well, I had hoped when I purchased this car that I would use it to break the outright electric vehicle record, which at that point was like 42 hours and change. And so this was the perfect car in which to do it. Unfortunately, Porsche released a new Taycan model that when configured rear wheel drive only with larger battery option and stripped of everything, is capable of more.
And right as I was preparing this vehicle.
Oh, my God, what's his name?
Right as I was preparing this vehicle to do it, Kyle from Out of Spec Studios went across an Ataycan In, I think, 3927, rendering this car obsolete for the purposes of electric vehicle speed records, but perfect for autonomous or FSD or semi autonomous records. So I'll have to wait for Lucid to release a new vehicle with a different battery architecture and then it's on.
[00:25:50] Speaker A: And Lucid's here in town, correct?
[00:25:51] Speaker C: In Phoenix, Yeah. Lucid has a facility in Casa Grande. Great company, Love the cars, their bottleneck. And this is, you know, typical of things people don't think about.
For internal combustion Cannonball or electric Cannonball or startups, the enabling technology may not be in your control.
So the enabling technology for elevators in the late, you know, late 19th century was alternating current power that could be transmitted over distance, such that you could build a skyscraper, which would mean you had a platform and demand for an elevator and Then an elevator that could go up and down quickly, repeatedly and reliably in the case of a lucid cannonball.
The limiting factor is the uptime and charge rate of the Electrify America network, which worked on the day that Kyle from Out of Spec did it. But I'm quite sure he had a friend on the inside of Electrify America to call and that is not something that's readily available for everyone.
[00:27:02] Speaker A: Well, this has been fantastic. What a spectacular car.
Thanks for watching Infinite Future. We will be right back in studio with more fantastic stuff from Alex. We'll be right back with more conversations at the edge of technology and transformation. Stay tuned. Every week on Infinite Future, we explore breakthrough innovation across every major frontier. We talk to AI architects, biotech pioneers, space entrepreneurs, clean energy disruptors and the thinkers redesigning global systems. We don't just talk trends, we examine scalability, ethics, economic impact and real world implementation.
If you're building, investing in or leading the future, join me on Infinite Future only on NOW Media tv. Because the future isn't predicted, it's engineered. And we're back. I'm Todd Thomas and this is Infinite Future on NOW Media tv. Let's look ahead.
Welcome back to Infinite Future. We're here with Alex Roy.
Thanks so much for the tour of the car that was spectacular.
Now last week I had Dr. Alan Badot on the show and he is an expert in agentic AI and he was talking about a bunch of different AI technology and different ways we can use it to optimize for performance across a bunch of areas. One of the things I found really interesting was agentic AIs that basically learn their human and become more responsive and more productive in interacting with that human.
Do you see a place for that in the functioning of a hands free vehicle?
[00:28:46] Speaker B: I do.
I fear that it will not come soon but but I am certain it will come.
[00:28:54] Speaker C: So if one looks at the driver
[00:28:58] Speaker B: assistance or any form of partial automation in any vehicle you can buy today, it is all designed according to serial automation theory. So you have a set of automatic emergency braking, traction control, lane departure warning and in some cases L2 systems have have lane keeping.
And then that lets Tesla Autopilot and the current version of Tesla fsd. But all of these are, they're serial automation in the sense that they have the technology, the automation has a zero sum relationship to human input. So as I engage more systems and it does more, I am doing less or I am doing something else. I'm monitoring the system, I'm merely a supervisor. So a Human in the loop, but absolutely doing less actual decision making about speed and direction. And so that is absolutely the end state of that is driverless and the guaranteed end state of that is skill decrement for human drivers, because there's no opportunity to learn. And so all of those systems are analogous to what you would see in Boeing for the entire history of Boeing.
So there is a schism in the philosophy of aviation automation. Boeing and Airbus have been fighting about this since 1980 or so. So if you get into, you fly any Boeing, even Boeing's automation, they have a variety of safety features and partial automation, but all of it functions the way driver assistance does in vehicles. In an Airbus, the system is completely different. You have input and you have output, but you have a software layer called the flight envelope protection system. And so given any altitude, speed, air pressure, whatever, the flight envelope protection system will, no matter what input the pilot puts, the system will boundary the outputs.
This does not occur in a Boeing. In a Boeing, if you are to, if you do see you're flying, you are pit nose down, it will, an alarm will go off and there will be haptic feedback. The pilot can ignore it, but in an Airbus, it will not allow it to happen.
Interestingly, despite the two approaches being completely different, the crash rates are the same, the reasons for the crashes often differ, but the rates are the same. So this isn't right or wrong, but when applied to ground vehicles, and especially personally owned vehicles, the opportunity to put in an Airbus type system, call it a driving envelope protection system, that's holistic, the way Airbus does it, we have not yet seen anything like it. And that opportunity presents an extraordinary potential market decades from now. And I'll explain what that looks like.
If one's sole goal was safety, then logically you'd want to increase automation, zero sum serial automation, and get to driverless vehicles according to the current path that we see for ground vehicles. But if you want to preserve the intrinsic cultural and emotional value of private car ownership, private car ownership, you must eventually deploy a parallel automation system like an Airbus for cars. Because all vehicles, all vehicles, personally owned vehicles, are sold to fill one of two needs.
Have more sex.
Do I love my family?
If it was just transportation, no one would buy anything but a Toyota Prius.
But the existence of the performance division of every oem, M, Audi S, you know, Mercedes, amg, the existence of Porsche as a brand, a Ferrari as a brand, prove what I'm saying is true.
Conversely, the existence of Volvo as a brand whose only message was do I love My family proves it's true. Now, for the last 30, 40 years, these two messages in marketing and two demand pulls among consumers have merged. Now you see Porsche has an suv, Ferrari has an suv. You could buy an SUV from a traditional love my family manufacturer with a sport package. Everyone's trying to be all things to all people. But what autonomy does and what electrification does is eliminate any differentiation in the product other than brand.
Eventually, when every vehicle is electric and it's inevitable on some timeline, it's like in the movie Fight Club.
On a long enough timeline, the survival rate of everything is zero. But in a long enough timeline, the success of all technology is guaranteed. Guaranteed.
So in the case of autonomy and electrification, there is no differentiation of actual product other than brand. And whatever design, exterior design flows, follows that brand.
Unless.
Unless Porsche or Ferrari or someone in the future goes back to the roots of marketing and messaging transportation products. Because a waymo could be 100% safe, perfect, it could be free, and I would not want to use one unless I also owned a car for those times I don't want to drive or I don't want to be in my car, or I don't want to park my car somewhere, I just want to get there, not think about it. But a Waymo disintermediates something very valuable, which is the need of a human, I'm not going to call it have more sex. The need of a human to project onto the vehicle they buy, their self perception and the need of that same human to be perceived by the audience of the world for being in that car.
AVs eliminate that.
There needs to be a substitute. The substitute is only personal car ownership, which is why personal ownership will always survive.
And autonomy will have to be an option on those vehicles, culturally and politically will have to be, because eventually autonomy will be without question proven to be safer than human driving. And insurance rates will reflect that. Making personally owned vehicles that are human driven almost unattainable. Unless, and this is where the product planners of the future will glue this all together.
Unless I can have a personally owned vehicle that has driverless functionality, that looks, that is a Porsche, and most importantly, and here's the agentic AI component that you asked about that makes us full circle.
The final piece of projecting onto the world, my agency as a human driver is them seeing me not only in a car that can go fast, but to see my hands on the wheel appearing to drive it.
That will become much like horse ownership today, a key part of that necessary projection for the person who is projecting. I can afford this car, I am in this car, my hands are on the wheel. But for that to function, for that to even be available, it must be productized. And the productization will be. After autonomy is ubiquitous and demonstrably safer than human driving, it will become possible to implement an Airbus type driving envelope protection system inside the autonomy envelope. This is what the vehicle can do inside that envelope in real time. The onboard vehicle AI will determine to what extent the allowable performance of the car will expand or contract based on you.
And so you can drive the car, hands on wheel, inside the envelope of safety determined by the car. And as you demonstrate over the time that you can drive the car closer and closer to the boundary of what the autonomy stack determines is the limit, you will be allowed to get closer, much closer. But if you demonstrate repeatedly that you cannot, that you're driving towards a group of kids, the system will bounder you very tightly. Your insurance at that point might be real time variation correlated to your inputs, but even though the outputs are boundaried,
[00:37:29] Speaker C: because of course there's always the edge case.
[00:37:32] Speaker B: So I am super optimistic about agentic AI implemented in real time in personally owned vehicles with autonomy functions, with a product that, that I'm willing to pay for for a subscription, which is precisely that. And I'm quite sure that every performance brand on earth will offer precisely this in our lifetimes. But it'll only come after a critical, not even the critical mass ubiquity of autonomy is real.
[00:37:59] Speaker A: Wow, that was fascinating. Thank you so much.
I'm Todd Thomas, this is Infinite Future. We'll be right back with one more segment with with Alex Roy. Please stay with us. We'll be right back with more conversations at the edge of technology and transformation. Stay tuned. Every week on Infinite Future we explore breakthrough innovation across every major frontier. We talk to AI architects, biotech pioneers, space entrepreneurs, clean energy disruptors and the thinkers redesigning global systems. We don't just talk trust trends. We examine scalability, ethics, economic impact and real world implementation.
If you're building, investing in or leading the future, join me on Infinite Future only on NOW Media tv. Because the future isn't predicted, it's engineered. And we're back. I'm Todd Thomas and this is Infinite Future on NOW Media tv. Let's look ahead.
Welcome back to Infinite Future. We've talked about what's real, what's hype and narrative shape around markets. Now I want to close with something deeper, the value question. Alex, your human driving manifesto argues that freedom of movement and an incremental Approach to technology.
Not anti technology, but pro human. And I think you touched on that a little bit in the last, really, I guess, addressing the agentic AI and the role of human in that.
What was the exact moment when you felt compelled to draw a line and launch the Human Driving Association?
[00:39:43] Speaker C: In, I believe is in 20, late
[00:39:46] Speaker B: 2017, early 2018, there was another school shooting tragedy.
And a lot of people of course, were talking about gun control.
And so I always follow the data and I think any reasonable person can agree that if a death is preventable, we should try to find a way to prevent it regardless of how it occurred. Can we prevent this?
And I've lost friends and family to illness and car crashes, all of it. I'm like, okay, so what are the most easily prevented deaths?
Car crashes. And you know, there is an entire political community around the second amendment and, you know, and gun ownership, but there's no political community, say, working on behalf of drivers. And there's the aaa, but they are not a political organization in the classic sense of what polity means.
And so if one deconstructs the messaging coming out of the autonomous vehicle community, it's very clear in the future, no one will own a car.
Every vehicle will be shared, they'll be electric, and they will be connected.
The acronym is case. Now, if you glue these things together, you're creating a big cactus to swallow and one that's going to take like 50 years to play out. That sucks.
Even if you want all those things to happen, even if all those things are good.
I believe in electric cars, okay? I believe in autonomy. Great. I believe in sharing when I feel like it. I grew up in New York, I took the subway. And I believe in, I guess connected to the extent that it needs real time traffic data for maps. But I am not interested in the future where I don't own the vehicle, where I can't drive it if it encounters something it can't resolve, where it's connected and invades my privacy and is electric because I have no choice.
This is a very American point of view, but it is a point of view that many people in the world want for themselves and it's why people want to come here.
That's a different show.
Now, looking at that, I'm like, wait a second.
There needs to be an organization that is fighting for the freedom to own a car and drive a car. But that organization would have no moral standing unless it absorbed and integrated the inevitability of automation and autonomy in vehicles in an interim phase during which we have partial automation. And in a final phase, the final phase to me is very clear. Autonomy is available in vehicles, but I can drive if I feel like it. Inside the safety envelope that autonomy defines, I'm cool with that. I own it. It's great. But the intermediate phase, the phase where language is vague and people don't understand what automation can do for them, or even if it works, that intermediate phase, there has to be one where education of consumers is core. And if people abuse their driving privileges, whether or not automation is involved, they should lose them. Absolutely, they should lose them.
And to me, the ultimate freedom is excellence. If someone's great at something, they should go do it to the extent that they can, as long as it doesn't interfere with anyone else. And. And so someone driving a vehicle who's not capable of driving it or understanding the automation on board should not be driving a vehicle.
I'm for a one strike and you're out when it comes to that. And so that was the genesis of the hda. I looked at the NRA manifesto. I did a find and replace guns, cars, shooting, driving. I integrated all of the core messaging points of the autonomy industry and eliminated everything that ignored human nature. And hit publish.
And within a week I had 10,000 t shirt orders, thousands of people signing for the mailing list, and interviews all over.
Which is funny, because to me it was only ever meant to be the spinal tap of automotive political action.
And I realized I had touched a nerve. And that is that nerve that guides everything I think about as I see press releases for any startup that involves robotics, because there's no shortage of robotic startups. But are they actually deploying the right automation in the right place, at the right time, at reasonable costs? Almost always, no.
[00:44:32] Speaker A: Well, that seems to be the general argument for AI and robotics in general. People are launching AI projects, they're launching robotic projects for the sake of launching a project, where I think, as you and I would probably agree, if you're launching a startup, you should start with the problem that your startup is trying to solve and let that problem define the solution, maybe it's an AI solution, maybe it's a robotic solution, maybe it's something completely different. But if you define a solution based on the problem, where you find real market need and people will pay to solve that problem, well, now you have the chance of a successful startup to build a robotics program or to build an AI just for the sake of doing it. That's why we're seeing ridiculous numbers like 90% of AI projects failing, because they're not starting with a problem.
[00:45:20] Speaker B: Yeah, it is fascinating what you just said, because the creators of the models made it possible, or they enabled hundreds of thousands of startups to build wrappers and then market those to like sub verticals and then kneecap them every 12 months.
And, you know, and so, you know, I don't, I mean, AI as, even as a concept is most akin to electricity.
But what happened with electricity is, you know, Westinghouse and these companies, they did not choose to then go into the business of making every single product that required electricity. And the difference now is the AI companies are in fact creating some products using their own foundational technology, which makes it very dangerous for anyone else. Very, very dangerous.
[00:46:21] Speaker A: It does.
So a few miles from here is the McCormick Railroad Park.
And they've got these cute little trains that go around on tracks and they've, they've got a couple of steam engine trains, they've got some diesel trains. So there's this cool old park that if you want to experience real trains, although at a different scale, but if you want to experience real trains, you can go there and experience in them.
Do you see a future where similar type setup for ICE vehicles?
[00:46:54] Speaker C: Well, the, you know, traditional ICE vehicle,
[00:46:58] Speaker B: I think a life cycle is like 11 years now.
And as long as they're sold, just take the year in which they are no longer sold and add 20.
And so over time, parts will come more expensive, insurance will become more expensive, they'll cycle out. And so the end state, I suppose, would be a park that has ICE vehicles. But, but really most ICE vehicles aren't worth.
They're not that fun to drive, and so they're not worth keeping.
We have automotive museums and some of them are attached to racetracks and we already have vintage car races.
So some variation of what you suggest is likely to occur.
I actually think we're gonna see something else emerge or become more popular to supplement that concept.
Eventually, every electric vehicle commoditized skateboard will have a near infinite number of top hats you can place on it. So I'm Ford or Porsche or anybody, and people are saying today my brand equity is going to go to zero because all performance commoditized. I'm like, actually, no, because in the future, Ford can sell a skateboard and you could put any top hat you want on it, meaning any, any generation of Mustang body can go on it, but it could be a new one. But the real value is in a database of the sound and physics of every Ford Mustang ever made. And a subscription model Like a video game where I can subscribe to download the noise and physics model of every Mustang ever made or every 911 ever made. So I can have every. I can own every 911 ever made and I can just select which year and the noise and the physics of that year.
And then the electric skateboard will replicate those physics.
And so. And if I have enough money, I can swap the top hat and then I can keep my actual collector ice vehicle at home and pull it out only on the track. But for my purposes, I can drive the experience and for, you know, projection, I can drive around with the replica top hat and that will be.
That will resolve like 90% of the demand to see old vehicles in motion and while eliminating the risk of driving those vehicles anywhere. And then there'll be a park like a racetrack will be like a machine park. It's a machine zoo is really what you're describing. And yeah, we'll have those.
We'll have those. It'll probably be a hybrid where you have robot fighting.
[00:49:22] Speaker C: Robot.
[00:49:23] Speaker B: You'll have robot fighting at a location. It'll be like a machine park with a zoo. You can see all the old machines do their thing. Some will fight, some will race. You can drive them maybe, and we'll have it all. I mean, I'm super optimistic about the future in almost every dimension because the future is a choice and people could choose really bad technology. It's happened.
They could also choose great technology.
[00:49:49] Speaker C: So we have to vote with our pocketbooks.
[00:49:53] Speaker A: Well, thank you so much for joining.
Just couldn't be happier that you're here. A true pioneer and I would say futurist. Certainly a systems thinker, but just really thrilled that you could join us. Thanks so much for taking the time. If people want to follow you or get in touch with you or. Or follow your cannonballs, where can they find you? Where can they follow you?
[00:50:12] Speaker B: LinkedIn Alex Roy x AlexRoy144 Instagram AlexRoy144 I'm easily found.
[00:50:23] Speaker A: Now, one last thank you to you if you are a attendee of ces.
Every year at ces, Alex throws the event of the year, the Atono Cast party.
Thank you for including me in years past and keep throwing that party and
[00:50:43] Speaker B: I can guarantee you will get in every year. I don't know if I can accommodate all of your listeners.
Thank you.
[00:50:50] Speaker A: I'm sure you can. I just want them to be jealous when I shoot a show from the party.
[00:50:54] Speaker B: You may.
[00:50:55] Speaker A: Thanks so much. You're watching Infinite Future. I'm Todd Thomas, thank you so much. We'll see you next week. Thank you for watching Infinite Future. I'm Todd Thomas. Join us next week as we continue exploring the ideas and innovators building what's next only on NOW Media tv.
