Can a driverless car really cope with London traffic & is the big AI bromance over?

0:03so danny we should say for listeners who can't see us we've both turned up to the recording of this podcast today looking quite sad well just like a couple loser tech reporters show me your t-shirt sit up a little bit you've got google cloud blazing across your chest yeah and you know what's cool it's not just any tech swag this is like extra soft cotton okay don't be jealous okay I would be jealous, but I can go one better because I've got, can you see this? Oh, what you got there? NVIDIA, GTC, Paris with a picture of the Eiffel Tower and the Seine. And in a sort of Monet style with GTC and large letters at the bottom.

0:49So very sad. It's like a concert shirt, you know, like, oh man, I saw Jensen on tour. I saw them in Paris. No, I saw them in San Jose. I did look on eBay and these GTC t-shirts go up to about 80 quid. You're not serious. Well, I'm obviously not going to sell it because it's too precious. I mean, obviously.

1:15Hello and welcome to the Times Tech Podcast with me, Katie Prescott. And me, Danny Fortson. Coming up today, we talk about AI talent wars and the odd, you know, $100 million salaries being thrown around. What's$100 million between friends? Plus Anthropik's big win, kind of, over use of books and copyrighted materials and training their AI models. And I go on a drive, but without a driver, but with the boss of the driverless car company, Wave in London, to find out if this is the end of driving as we know it in the UK. But first, Katie, the story you wrote about this very week is the bromance, this famous bromance between Sam Altman and Satya Nadella dead over kaput.

2:05Or as you put it more kindly, under pressure. I think that might have been subbed by somebody else. But anyway, yeah, this is an absolutely amazing story. If you think of the relationship between Microsoft and OpenAI, the strategic partnership, as they called it when it when it started back in 2019, and Microsoft put a billion dollars into OpenAI, which, you know, turned out to be quite a good bet, you might say. At the time, they didn't get a stake in the company, partly because of this very strange structure that OpenAI had as a not for profit. But they did get a share of the profits, albeit a capped one.

2:43And now OpenAI is fiddling around with its structure and has taken on a lot more funding and it's got people like SoftBank involved and their valuation is 300 billion. They need to renegotiate their relationship with Microsoft. And so, as you say, this bromance between Sam and Satya is suddenly a little colder than it once was. And so, but what is that? How does that manifest itself? Is it like a whisper campaign? is it's like you know like oh these things are leaking out and there's some interesting well-sourced stories about how sad to be. Yeah that's exactly it. This is no Elon Musk fight with Donald Trump.

3:23This is absolutely leaks behind the scenes about what's going on in these negotiations but I think the most fascinating thing about it is it all bizarrely hinges on AGI right? The definition of. It's the definition of it but also it is a sort of nebulous concept which I find really interesting because as we know sam altman says that artificial general intelligence is just around the corner and it has a definition for it which is a highly autonomous system that outperforms humans at most economically valuable work okay following yeah it had this fascinating clause in its contract with microsoft which basically said once it hit this point and once it got to AGI, Microsoft would no longer have rights to OpenAI's stuff.

4:14Or anything once it hit that point beyond. Like in other words, because from memory, Microsoft basically is part of their investment, which is why Satya Nadella has such leverage is that like they have free reign to do whatever they want with OpenAI's code, with their agents, with whatever. Exactly. Up until they hit AGI. But then what happens? What happens when we hit this magical point? So the contract says it can stop sharing new models with Microsoft. And obviously that's really, really problematic because people will want the new models because they will be AGI, they'll be the most advanced.

4:52And Microsoft's value from OpenAI, as you say, is in having access to all these models. It didn't have a stake in the business, but it did have access to the models. And so now they're renegotiating that relationship and Microsoft wants a stake in OpenAI in the new world of OpenAI. Right, right, right, right. And it wants to can this AGI nonsense. And it's really interesting when you look at what Sam and Satya have been saying about it. Sam Altman always says it's around the corner, we're on the cusp of it. Whereas Satya Nadella back in February rather poo-pooed the fact that it was coming. He said, us self-claiming some AGI milestone, that's just nonsensical benchmark hacking.

5:33I love it. Well, because of course they're both, they're both talking their book. You know, he's like, AGI, that's never going to happen. What's that? I get access forever and ever and ever. But it is interesting, like when Elon Musk sued OpenAI over their conversion from a nonprofit to a for-profit and him kind of going to war with Sam Altman in his lawsuit, he said, look, they've already reached AGI. And so he was trying to say that point has already arrived. If you look at the capabilities of these models, which, you know, is a stretch. I mean, they're helpful, but they're definitely not like the super intelligence that's better at humans than everything.

6:11That is not true. The important thing is who gets to decide. And in the clause at the moment, it's the open AI board. So you can see why Microsoft isn't that happy about it. It's interesting. And I guess ultimately, I think you're going to have to come down to some type of benchmarking, which is why there's, you know, there's this thing we mentioned before called humanity's last exam, where this organization has brought together like the top experts in all these different fields come up with all these really complex questions and said, when AI can basically ace this test. that is when they're calling it humanity's last exam because then you're like oh okay well we've been surpassed by the machines and it's this test that is kind of constantly being kind of built up and involved as they get more experts from around the world to come up with their biggest most naughty questions in their given field but i think that maybe it'll be something like that that ultimately defines this i suspect in the in the new contract they'll try and just get rid of it all together it's yeah as you say so hard to define and unless it applies to everybody else all the other partners yeah it doesn't maybe make sense in this in this new ai world but it does feel like open ai is soaring above its master which is quite funny yeah it is it's a feels like almost like love go on that's that's how beauty is in the eye of the beholder like somebody here you're just like you know if i'm the creator of this thing i'm like yes this thing's amazing it's it's like it can do everything and you know somebody else might be like meh not for me yeah you know yeah and i get the impression microsoft has tried to fall in love with other companies but no one's quite special exactly exactly um speaking of uh dating around and trying to land new partners in this crazy eye world our mate mark zuckerberg he's been throwing his money round obviously he spent 14 billion dollars on scale ai which is this company that trains helps train ai models and label data and all this stuff but there's been the story doing the rounds that he's been offering nine figure pay packages to ai engineers have you got the call weirdly no i mean no i did send him just cold emailed him my meatball mania video game that i vibe coded with your bank details at the bottom yeah i'm just like don't you don't even have to respond just like direct deposit my dude weirdly that hasn't happened yet i'm checking like i'm checking my bank account like every 15 minutes because apparently he's trying to work out on his list like how people like being approached you know does danny forlson prefer a whatsapp or an email i just prefer the money do you and then you'll turn up at nine o 'clock on monday No, I'm in demand.

9:10Have you seen my video game? I'm like, I just, I sit here and work my magic and you pay me all the money. Sam on one phone, Sasser on the other. But anyway, so yeah, so I haven't got the call from Mark, but he has managed to poach apparently at least four senior open AI engineers with these offers of, yeah, apparently he's offering up to a hundred million dollars to people. which is just clearly the, all of these guys think this is the biggest thing since sliced bread. Because if you're like, well, if I'm offering a hundred million dollars to individuals, but the prize is trillions, it's a drop in the bucket, of course.

9:56And it's like, there's this whole race mentality. Who's going to get there first? Who's going to build the best model? Who's going to kind of steal a march on everybody else? Open AI clearly is in the lead, but it's really quite extraordinary. And it makes the dot-com boom, which when I started my career out here, look, I mean, the numbers and just what's happening here are so insane. If you had a time machine, do you think it would be sensible to go back to your university days and maybe do a PhD in machine learning, swerving the humanities?

10:32Of course it would. And you can tell yourself. yourself. Well, we've chosen the wrong career, but I kind of knew that at the time, but I'm like, I'm just, my mind does not, is not quantitative like that. But even like, for example, like Mira Maradi, who was another open AI founder who left, she just raised$2 billion for her startup, $2 billion, no product, no revenue. It's like a PowerPoint deck and her and some smart people from open AI. That's like your starter. That's like your series a or your seed round is 2 billion dollars you know and all most of that money i'm sure is just going to get recycled right back into all the big hyperscalers which is why like microsoft and amazon and google are doing so well because everybody's just like taking all the money they raise and then paying for space in their data centers to train and run all of these models um so it's a little bit like of a washing machine you know and any money is left is just spent on the engineers on the people building these systems.

11:31I guess the slight danger as well is that you do end up consolidating the talent in the hands of the few again. I was speaking to one of the AI unicorns in London about this and they laughed when I said, were they thinking of paying those sorts of salaries? Of course they did because nobody else can. But it just means that maybe, unlike in the past where you take a stake in the company and then you'd hope it would grow and you'd make your future wealth that way, you can just go straight to meta now i'm just so waiting for my my email from mark hold on let me just check maybe it's in your spam while you while you tee up the next news item i'm just gonna see if mark has texted me check check your facebook messenger as well oh yeah because i know he forces out no still nothing from my guy he's weird he's busy don't worry just keep keep holding onto your phone.

12:25So I really, really want to talk about copyright as our last story. And that's partly because there's been this huge case going on here in the high court between Getty Images and Stability AI, which finished on Monday. And has there been any news out of that? Because I know that was like the big one, at least over on that side of the pond. Exactly. Yeah, it is. It is really the big one and the only one as the government here grapples with what to do about this copyright issue. So Getty Images alleged that Stability AI was using their material in their machine, in early versions of their machine, and that you could thus generate images with the Getty watermark on, including pornographic and inappropriate images, which they said damaged the brand.

13:08So they dropped a number of bits of the case as it went along, and it ended up being about really their trademark by the end, the Getty trademark, and what the damage that was being done to the brand. But it also really interestingly, and this is a bit geeky, but bear with me, they were initially looking at whether the models had been trained in the UK and trained abroad. They dropped the one that accused stability of training in the UK, but they kept the abroad one with the issue being that if the judge finds Fagetti, even if businesses have trained, AI businesses have trained their models abroad and then brought it into the UK, it could still breach somebody's copyright here.

13:48Because you're bringing in pirated materials. You're like, I used to live in Spain and you'd be like walking through the plaza and there'd be all these guys like selling like pirated CDs on a blanket, you know, then they could like pick it up in a moment and run if the police came. But it was like, you're selling stolen goods in another country, basically. That's exactly right. And it's so interesting you use that example because that's exactly what Stability have argued that it's for physical goods and not for things like AI. So the verdict's expected in a few months and that will give everyone a lot of clarity into the direction of travel over copyright.

14:27But there'd been two rulings at the same time in the US. Yeah, the big one was Anthropic and it was mostly a victory for them in that they were similarly being sued by authors who were saying basically you stole our stuff without our asking and you've used it to train your models and that is a copyright violation. And the judge said basically no, broadly speaking, no, with some nuance. So first of all, Anthropoc was like, they used 7 million pirated books. And that seems like not up for debate. That is a fact that has come out through the course of all these hearings. That still needs to be adjudicated and will be.

15:11What they then did is last year, they hired somebody from Google, the person who did their big books project, which was also very controversial. Google was wanting to put every book on the internet, digitize it and put it on the internet. So what they did under this guy's guidance, apparently, is buy millions of physical books, extract the pages physically from the book, scan them, turning each page into a digital document and then using that to train their models. And then afterwards they shredded all the books. Horrible thing to do. Yeah. But that buying of a book and then doing with, with it, what you will is why in this, in this ruling, he was like, basically this is fair use.

16:02You're fine. As long as you're using this as inspiration as long as you're purchasing these copyrighted materials and then doing with it what you will you're good the pirated books case which i think will that will continue but like if you just buy every book on the planet and scan it that apparently is okay according to this ruling wow um the i was speaking to a lawyer about this uh while i was covering the getty case and they said the damages that anthropic faces over these pirated books could be existential he said could be significant even existential which is really interesting but it is uh it is nuanced isn't it and it feels like everybody's trying to fiddle their way through it at the moment and no one's coming up with a satisfactory answer right let's take a little break then after the break katie you get driven on a driverless drive around london like toad of toad hall

17:09danny i've got a brilliant idea for your next date night when you come to london all right hit me so you could get in a driverless car and go around the streets of soho on a friday evening while the drunken patrons of the local theatres and pubs spill out onto chaotic single lane roads. That sounds incredibly romantic. Hopefully I'll see a couple of people barfing in the gutters. You're really lucky. But it's what today's guest likes to do, apparently. He's the boss and co-founder of Wave, the UK's answer to Waymo. And you can see if you founded a driverless car company, seeing your vehicles in action would be quite satisfying.

17:50but at the moment, unless you go on a date with Alex Kendall, the boss, I think this is going to be tricky to organize for you because we don't yet have driverless cars in London. I mean, it was already struggling. Romantic was doing a lot of work there, but now this sounds far less romantic. I don't think Mrs. F is going to go for that. Me, you and a random tech CEO, babe. Let's do it. Wearing my Google cloud t-shirt. I mean, but I, okay. Okay, romance aside, one of the perks of our job is obviously sometimes we get access to things that aren't yet open to the public. And a few weeks ago, I got to drive around London in one of Wave's driverless cars while talking to Alex.

18:32And it was pretty cool. Do you want to tell me a little bit about Wave with a Y? Because I think it's fair to say people know less about them than Waymo with a Y. Yeah. So Wave was founded in 2017 by Alex Kendall and another PhD student in Cambridge. And they had this idea that rather than train a car to drive along certain roads and follow a path, that it would learn through artificial intelligence by being fed information. And that's exactly what they did in Cambridge. So they had a little car, a Renault Twizy that looks a little bit like a golf buggy. And they worked on automating it and so that it could drive itself around the block.

19:20Now, fast forward to last year and SoftBank came in with a billion dollars along with NVIDIA and Microsoft. And Microsoft had been a previous investor in the company. And Wave is now one of the UK's top AI businesses. And it's recognized around the world. So they've now got offices in San Francisco, out in Japan, because of the links to SoftBank, in Stuttgart. And it's expanding at a rate of knots, largely due to this enormous bazooka investment from Massasson. Fascinating. Well, I want to hear about I want to hear about your ride. And then because, you know, we're living a little bit further in the future here because we have Waymo out here, which is everywhere.

20:04Yeah, it really is so unusual here. It was quite bamboozling as you'll hear. I'm interested to hear what the interview sounds like because I was so overexcited. I forgot most of my questions. So this was the second time I've been in a Wave car and I was lucky enough to go in one when they got that mega funding round last year. And just to set the scene before we play it, Wave's based in a large office on what is, it's kind of a scruffy road north of King's Cross. And it almost looks like it should be a garage or a petrol station. and we got into one of the cars which are all sitting out on the forecourt Alex and I sat in the back and there was a driver in the front right because at the moment you have to have a safety driver in a car in the UK and this is not an Elon Musk man with hands on his knees in the passenger seat this is sitting behind the wheel but they have their hands kind of cupped under the wheel and not touching it so if something happens they can step in.

21:04Hands on knees or hands cupped? Hands cupped. I'd love to make the laws around this. Great. So, so what we were in the back, there was a chap in the front, a safety driver, but he didn't have to touch the car once I should say. And we went down into central London. So if you think King's Cross is the North, we cut straight through into the center past the British museum and those little higgledy-peggledy streets around there, which were full of kids because it was half-term, and Great Ormond Street Hospital, which was unloading ambulances and all sorts of things. So it was a proper London experience, I have to say.

21:41Well, let's hear it.

21:44One of the things I found really scary about coming out in the car last time, well, scary is the wrong word, but kind of challenging when you're getting your head around being in a driverless car, is turning onto the York Road. It's not an easy road to drive on as a driver, let alone with the safety driver. Well, let's see. It's a busy road to turn on here. I always wonder, why did you choose to be in Kings Cross? Well, we were starting the company in Cambridge and as we were growing and looking for a space, ironically, we wanted to find a space that was really important to us to co-locate the car and our team so that everyone can go and test and get in the car whenever they want and see what the autonomy is like.

22:25And actually, Kings Cross, we found the best spot. in central london near where we all lived um with a warehouse that could house our team and our fleet because it sorry should we talk about the technology because you've just pressed the button to make the car autonomous is that's what is that what's just happened that's right we're sitting in here in one of our wave autonomous vehicles in central london our safety operator just turned the system on we've got a safety operator behind the wheel monitoring the system but the ai is now driving the car speed staring indicators the brakes everything about the car is now being driven by the AI.

22:57But my question about why Kings Cross was, I guess, also you would think that a car company might want to be somewhere more in the countryside with more open roads, with easier roads to drive on. Was the challenge part of the reason for being here? Since the beginning of WAVE, we've tried to take on the hardest problems first in building this because the autonomous driving industry many years ago started in places like Phoenix, Arizona, which is a city in the US and where there's a grid-like array of streets the sun away shines I think they have like one or two rainy days a year and one of the easiest places to learn to drive whereas you come to central London and it's medieval origins of cities with lots of merging and roundabouts and and of course the the weather that adds complexity here so coming here forced us to build something that would scale forced us to build a system that was intelligent to not just drive with protected red and green arrows on every intersection, but actually learn how to navigate and predict how others will move.

23:55An interesting stat, compared to San Francisco, where there are lots of self-driving cars, London has about seven times more jaywalkers. Is that right? The hazards are everywhere. Not to mention the line bikes, right? I mean, that must be an extra. Well, just now, we've already just driven through a pedestrian crossing with someone darting out from behind a vehicle. There's a line bike coming around beside us, and now we're just stopping here at a red light. So what's happening with the technology? What is the car? How is this working? How do you sum it up for people? Yeah, it's a core belief that we've had that autonomous driving is really an AI problem.

24:29It's a problem of how do we build a system that we can trust to drive through scenarios it may never have seen before because, of course, we can collect all the training data that we need in the world to understand the diversity of where we drive. But at the end of the day, you're always going to see things where you know how people and cars are on the road or the weather that it's always going to be different even if you drive down the same road twice and so the key thing here is how do we have a system that's intelligent enough to deal with things that may never have seen before and for us building an ai system was how to do this so since the beginning of the company we've been building a technology that learns to drive that learns the behaviors you might expect like here we're just slowing down for a speed bump we haven't told the car if there's a speed bump you should slow down instead it's learned from behavior and feedback that that's a policy it should learn it's learned how to go through traffic lights and that means that it can drive through any traffic light it sees rather than specifically just ones that have been mapped or or pre-coded how it how it should navigate through them i loved your stat about the number of jaywalkers compared to driverless cars because every time i speak to my podcast co-host in san francisco he talks about going in a driverless car i'm like you might as well be on a different planet because obviously we just don't have them here in that way yet or that experience.

25:46And I think the UK government has now said it's going to be 2027. That's right. So they'll be on the road. Last year, we were able to help the government pass the Automated Vehicles Act 2024, which puts in place the legislation to legalise autonomous driving here in the UK. Originally, it was meant to be enacted, the actual regulation implemented by 2026. It's been pushed back a year to 2027 and we now expect this to all come in place by the end of 2027. Because it's the second time I've done this. I definitely felt more comfortable this time. And I wanted to ask as well, do you feel like the tack has advanced in the last year?

26:25Oh, massively. I mean, a year ago we were just driving in central London. In fact, probably couldn't have done some of the complexity we'll see on the road today. But in the last year, we have learned to now drive on highways at high speeds. We now drive across the world in Germany, in Europe, in Japan, in Stuttgart. That was quite interesting. Do you see the pedestrian that was next to the crossing? A lot of self-driving cars will stop if there's a pedestrian anywhere near the crossing. But one of the things we got very good at is predicting from the pedestrian's body language. How are they shaping?

26:58Does their body language look like they're going to cross or not? Wow. So the sensors are looking at the people and how they're reacting. We've developed a world model. The AI is able to actually predict and understand how the world will evolve with very complex scenarios. So in that instance, the pedestrians' body language was not facing the crossing. They were just walking past. So we didn't stop. We just drove through, as you would. Whereas if they were shaping to cross the crossing, of course, we'd yield for them. And learning that kind of level of intelligence is really awesome. But over the last year, we've grown orders of magnitude and performance and then spread out to now drive all around the world from the autobahns in Germany to downtown Tokyo, for example.

27:36Actually, in the last 90 days, we've now driven in over 70 different cities, which is a pretty remarkable statistic. Because it's been a year since I saw you last, and that was on the occasion of you getting the billion dollar investment led by SoftBank. Is that what's precipitated so much of the technological change? We can talk about the company's growth in a minute. Yeah, I mean, that investment, of course, was a huge inflection point for us. Gave us the ability to build out everything required to turn this technology prototype into a robust product that can be trusted across the automotive industry.

28:16So there was a huge build out in turning this into a reliable product that can actually operate in consumer and commercial vehicles. And then, of course, a lot of the technology investments that we have been pioneering for the last eight years, I think, have just keep landing wins. You talked about pedestrians' behavior. Are there other things that have been really challenging to train the vehicle to do that you've managed to? Predominantly, the hard thing to do in this is multi-agent interaction is dealing and predicting how others are going to move. And actually, driving requires a lot of negotiation.

28:48I mean, consider if you're in a crowd of people and you need to nudge your way through or if you're in a merge situation. If you don't negotiate and if you sit back and wait for space, you might be stuck there for maybe an hour. Although I was in Tokyo two months ago when we launched our cars there for the first time. What was cool to see was a Japanese police officer jumped out in front of us and actually put his hand up and said, stop, wait, and wanted to let an emergency vehicle out from the side of the road. Our car, we hadn't had any specific training data from Japan yet, but from its general intelligence was able to understand to stop for that police officer.

29:25And then after they let the car out, then they waved us through and actually recognized that signal and then drove forward. And so the sheer general purpose intelligence that we're seeing emerging in this AI is quite remarkable and is very quickly proving itself to be able to drive all around the world. Talk to me then about how the investment has changed the business in the last year. So you've expanded now office in San Francisco, Germany, Japan as well. That's right. So there has been a year of global geo expansion. Of course, the investment's given our partners the trust to work with us because launching a product in automotive requires years of development and then years of expansion.

30:02And we have that business plan fully funded and the runway to actually be trusted in the automotive space. and of course continue to grow our team and it's no secret there's an AI talent war going on at the moment and the expertise that we've been able to bring on board and that I'm so thrilled to work with at Wave has been awesome. One of my favourite things to do is go autonomous driving on a say on a Friday night through Soho where you have all of the people coming out of the theatre very very busy you can see some amazing scenarios there and just the sheer intelligence of our system to safely navigate through those kind of scenarios when of course you have all the drunken people spilling out of bars and shows that that you might expect i can sit and people just walk out into the road oh there we go punchy confident yeah confident driving does the car have personality um something you can personalize you can prompt the ai and ask her to drive in the in the style that you might want so aggressive london shouty your words not mine but if you want to drive that way.

31:04Of course, our AI understands the envelope of what's safe and what the boundaries are. Oh, someone spotted it. There's a guy outside who's grinning at the car. Yeah, these things are still a bit novel in London, but I hope really soon everyone gets to experience. So 2027, this is the plan that in the UK we'll see autonomous vehicles on the road. So in 2027, to be specific, in 2027, it is the government's plan to have legal frameworks implemented so that you can get approval to run robotaxis or driverless vehicles. We have not yet publicly shared our date that we'll be launching such a capability, but we're working on it as quickly as we can.

31:46And we're really excited to do so with that regulatory approval. And have you got a sense, and we don't have to talk specifically about WAVE, but when you think driverless cars in London might become so common that you won't see that chap stopping and going, hey, has that man got his hands on the wheel? As soon as they're legal, I mean, as I said, We'll be working on this as quickly as we can. We've got some great partners from the automotive companies that we work with. Of course, Uber, who's a really great partner of ours. And we're working together on as quickly as we can bringing this technology to the world and doing it in a way that's safe and scalable, which may take it beyond some of the limited deployments we've seen so far in cities like Shanghai and San Francisco.

32:23Just to be clear, so Wave works. Your technology is then embedded in partner cars. So, sorry, with car makers. Essentially, you've just signed a deal with Nissan, for example. That's right. So we're building the embodied AI platform across different manufacturers and operators of different robots. So whether this is a car company or other form of mobility service or manufacturer, we've built a foundation model that's really good at understanding how to navigate through complex scenes, how to drive safely, how to understand the physical world. That's embodied AI, AI in the physical world. and this foundation model is general purpose so we can optimize and deploy it in a driver assistance application in a consumer vehicle or we can deploy it in a robo taxi or in another form of robotics and this is this platform that we want to partner with the world's most innovative fleets and automakers to allow them to create autonomous products so first that'll be with driver assistance products then robo taxi and autonomous mobility and then of course many other autonomous machines that can make our lives more safe and sustainable what are your thoughts on where we'll be with car ownership and something like gosh this bike just crossed over he's not wearing a helmet but in terms of safely responded and it did yeah yielded for them because in london it feels to me like car ownership is declining do you see your technology in cars that are going to be used by organizations like Uber or do you think people will continue to own cars?

33:57I think that's a really interesting question. The first thing is I think multimodal transportation and different options is going to be really important to people. So I think it's interesting to argue about the distribution across different modes, but I think there's going to be a non-zero amount of every mode and that's important. Trains, bikes, walking, micro-mobility, public transport, shared transport, personally owned transport. I think we need an arrange to meet the diversity of people's needs. So that's the first thing I think is really, really key here. I cycle to work every day and I think, I hope cycling is still a key part of the way we get around.

34:33Yeah. But I do think that autonomy provides a number of opportunities. You know, think about when you improve road safety, what that lets you do for transportation, improve accessibility, reduce the cost, of course, by making vehicles more highly utilized. Today, cars sit idle on average 97 % of the time. Is that right? Yeah. And so being able to utilize these vehicles better, to be able to run them more efficiently, I mean, that opens up many more applications, whether this is lower volume, more frequent public transport, or maybe even personalized public transport on the route. I think we will see an increase in ride hailing and shared car ownership.

35:12I think that's natural. Imagine if your Uber, you know, costs a quarter of the cost that it costs today right i mean i think that would be if a you know 30 pound or 30 uber was uh all of a sudden 10 or 15 pounds or dollars that really does swing in favor i think uh the availability of that transport that can be made more efficient and then even like how much space we waste with car parking in in london 10 times the size of hyde park is taken up by car parks and when you have cars that aren't sitting idle but are constantly utilised, you can turn those car parks into green parks and make that space more efficient.

35:49So I think there's a ton of changes we are going to see the way we move around cities, not to mention by reducing accidents. We can, of course, reduce congestion and make actually the journey faster and more effective. Do you think that it's likely that autonomous cars powered by waves technology are going to be on the roads in other cities before they are in London? I mean, is it easier for you to deploy them in San Francisco where they're already? We haven't committed to which city will be first yet, but I can comment in the general sense. I mean, London, we've got maybe home turf advantage here.

36:22So there's an attraction on that point. But we are seeing that the regulatory landscape in the US is more advanced. We are also seeing automotive manufacturers in general in Asia and Japan and China and the United States are certainly further ahead than those in Europe. So I think there's going to be some interesting factors as to how this technology goes out. I think, of course, market size, regulatory landscape, consumer adoption and acceptance, as well as the manufacturing and operational support. All of these things are going to be important factors. It's unclear, but I certainly want to make sure that London is one of our top priorities.

37:00Do you still get a buzz when you're out in the car? Every time I'm in the car. I mean, isn't this magical? It's completely magical. And it's so weird to think how many years ago you were just in your little tizzy car in Cambridge. The Reno Twizy. Twizy, sorry. The prototype we built in the garage of the house where we started off. This was in Glisson Road in Cambridge. And yeah, we taught that car to drive around the block, but there it was every 100 meters needed some human disengagement. And the fact that we've driven for about an hour now around London with some of the most complex scenarios you can imagine.

37:31and our autonomous AI is driven through them with no human disengagement or interaction and completely autonomously, that's been magic. And even right here, we just got through a double roundabout. With a diversion sign, a road closure sign to our left. Two roundabouts literally touching each other. So it's like a figure of eight. We just had to drive through one and then the other. I mean, not quite the Swindon roundabout, but that's pretty special. It reminds me of learning to drive a bit where you just suddenly start noticing all of these things. I failed my first driving test I should say in New Zealand.

38:03Did you? That's really interesting. I'll use that, thank you. I think I ran through an amber light. Did you? Yes. That would inspire you to start the business. I wouldn't say I'm a terrible driver, but I did pass my second time. I've never actually owned a car, except my company has a fleet, so I don't know how you define that. But yeah, to be honest, I get bored by long car journeys. I would much rather use my time for something else than spend time behind the wheel. And so being able to build autonomous systems is something that would please me as a customer. How are you dealing with all the pressure on you at the moment?

38:44Because I see Wave everywhere. You're like the great British AI hope in so many ways. And people talk about you as, you know, you are the leading British, one of the leading British AI companies. And your life must have changed immeasurably, you know, since you founded the business, but also now with this huge investment you've got and growing globally. Yeah, how are you kind of mentally shouldering that? Well, thank you for that. I think we're still very early on in our novel, only in the first few chapters. And I found this just to be a whole adventure, right? every year is just presented new things like this year traveling around the world working on multi-year long deals with major car manufacturers that are worth billions of dollars to bring cars to millions of vehicles I mean this is something that even just last year of course building up the technology to be able to drive safely on on these roads like every year brings a new challenge so I love that challenge I I love uh I love the learning and growth that comes with it it feels like whenever I get good at a skill or comfortable with the process we outgrow it and I love living in that zone of uncomfortableness and I also think there's been a proportionality to the growth I mean it hasn't just happened overnight this has been eight years even longer during my PhD as well you know over my whole life I've been working on robotics so it's been been many years in the making you know we let our results do the talking and I think I'm I'm really proud of being out for us being able to drive in for an hour under autonomy throughout central London like it's a really proud moment we do show this with more confidence now because we feel like it's ready to bring to the world and and you know feel confident about the evidence and you know capabilities that we have that it can actually bring those benefits to the world it's not adding hype or you know false promises but genuinely this is something that will change the cities we live in does not have the limitations of the earlier iterations of autonomous driving technology but is something how they can drive worldwide with the level of intelligence that can actually be safe and useful we're back we're turning back into into your offices amazing yes here we are back at wave headquarters in king's cross just finishing some park line bikes in the middle of the road and a lorry that's unloading one more negotiation just if you hadn't enough had enough but here we are back at the office and um and that means that beep means we've finally disengaged autonomy after an hour of driving through central London and here we are we'll just manually drive it back into our garage and get on with our lives you know when you said one of your favorite things is driving around Soho in the evenings did you do that sometimes just for fun yeah yeah um it's uh yeah sometimes Friday night spin Friday night date night uh cruising around in the autonomous vehicle.

41:38That's pretty cool. Is it the future? I think it's the future. I think it's incredibly impressive. It is the second time I've been in a wave car. As I said in the interview, it felt like the technology had come on leaps and bounds. Alex seemed to feel more relaxed about it as well. It's very surreal. And I know you're used to this. So forgive me for getting a bit overexcited about it. But it is very, very surreal when you sit in the back of the car and you see the steering wheel moving by itself. It's the first time the steering wheel moves. It's like a visceral experience, right? Of like, oh, whoa.

42:16And then seeing it navigate these incredibly complicated situations. And it was funny being in the car with Alex because it was nice to sit next to a driving instructor, you know, when you're learning to drive. People are like, oh, look, oh, watch out for your lap. Oh, there's a cyclist coming up. Make sure it comes out the other side and all of that stuff. So he's looking at it. all of these hazards he's watching all these hazards all the time but through an ai lens so he's a combination of driving instructor and machine learning phd so he gets geekily fascinated as you could hear by like the problem the conundrum posed by a double roundabout yeah but it was very impressive to see the car navigate it so confidently because london is not an easy place to drive in and if you think of all of the problems that you could replace with driverless cars people drinking people being on the phone you're so many like this is a terrible admission i had to do a speed awareness course recently do you know what that is basically you got caught you got a ticket a speeding ticket and now you got a speeding ticket i didn't wasn't going very fast i was saying something like 24 in a 20 zone to be fair to me i don't drive very often in london but anyway i did get caught and so to avoid getting points in your license you do a speed awareness course and the driving instructor was going through i think it was actually a policeman saying what are the things that make you drive too fast it's being angry it's being anxious it's being stressed how can you mitigate those circumstances you know like well if the car's driven by a computer yeah i also sometimes drive fast when there's a good song on and i'm like rocking out do that though no it wouldn't it wouldn't look danny before we go there's one last piece of major tech news that's happening and it's happening today oh yeah do tell well the time stick podcast has has a terrific producer called callum mccray and he's uh yes he's leaving It's so sad.

44:16It's so sad. Cue the sad music. Callum, as your last act on this podcast, can you cue in some sad music for this next bit? This is a sad day for the Times Tech podcast because you've basically launched this and I was doing the Danny in the Valley thing on my own out here in the wilderness for years on less than a shoestring. So to have you come in and be like a total pro and actually make us sound way smarter than we are has been a total just an amazing experience for me and we're very sad to see you go no we really are you've been absolutely fantastic and and we will miss you very much yeah you've made this thing what it is which i totally appreciate and i'm sure i know you do too katie yeah absolutely wish us luck listeners on our journey without yes yes our calumnless our calmness existence.

45:08It's going to be tough. Maybe we can turn him into an AI.

45:37Thank you.