WEBVTT - Self-Driving Cars Might Finally Be For Real This Time 0:00:10.200 --> 0:00:13.640 Hello, and welcome to another episode of the Odd Lots Podcast. 0:00:13.720 --> 0:00:16.000 I'm Joe Wisenthal and I'm Tracy Alloway. 0:00:16.640 --> 0:00:19.680 Tracy, remember the hype about self driving cars from like 0:00:19.680 --> 0:00:21.320 ten years ago, Like that really died out? 0:00:21.640 --> 0:00:24.279 Can I tell you something? Yeah, I'm still holding out 0:00:24.320 --> 0:00:27.560 hope for the self driving cars because I can't drive, 0:00:28.160 --> 0:00:31.080 and it was kind of I forgot. It was kind 0:00:31.080 --> 0:00:34.080 of acceptable when I was in my twenties, but now 0:00:34.120 --> 0:00:36.400 it's starting to get a little embarrassing. So I really 0:00:36.440 --> 0:00:40.040 need the self driving cars to become viable options. 0:00:40.200 --> 0:00:43.080 So when we go on the road and like do podcasts, 0:00:43.200 --> 0:00:44.120 like in another. 0:00:43.920 --> 0:00:46.199 City, you have to drive. I have to drive all 0:00:46.200 --> 0:00:48.360 the time, don't It hadn't because I think I've asked you. 0:00:48.400 --> 0:00:49.960 I'm like, oh, Tracy, you're renting a car. 0:00:50.159 --> 0:00:52.159 And then you like sort of change the discussion or 0:00:52.240 --> 0:00:54.280 you bring up something else, like oh, I think there's 0:00:54.400 --> 0:00:55.800 Ubers in that town or something. 0:00:55.880 --> 0:00:57.320 But this is the real reason, isn't it. 0:00:57.360 --> 0:00:59.040 Well, I mean there are Ubers everywhere, you know. What 0:00:59.080 --> 0:01:02.040 we should talk about whether or not Ubers have decreased 0:01:02.200 --> 0:01:04.880 enthusiasm for self driving cars as a business model. 0:01:05.400 --> 0:01:07.560 Remember when people used to say Uber would never make 0:01:07.600 --> 0:01:09.640 any money if they still had to pay humans, but. 0:01:09.560 --> 0:01:10.679 Now they're making a little money. 0:01:11.080 --> 0:01:11.720 It's true. 0:01:12.280 --> 0:01:15.119 But I do think like generally, like when people talk 0:01:15.200 --> 0:01:17.759 about like tech that didn't live up to the hype 0:01:18.400 --> 0:01:20.440 and now you see it now with like chatbots and 0:01:20.480 --> 0:01:22.720 stuff like that, and whether they're legan chain like people 0:01:22.760 --> 0:01:24.760 go back to the self driving cars, Like to me, 0:01:24.880 --> 0:01:28.520 that's the sort of quintessential example of the like modern times, 0:01:28.760 --> 0:01:30.920 maybe three D printing. You don't really hear that much 0:01:30.920 --> 0:01:31.520 about it. 0:01:31.520 --> 0:01:34.000 Yes, but don't you also find it weird to imagine 0:01:34.040 --> 0:01:36.880 a future in two or three hundred years where there 0:01:36.920 --> 0:01:40.520 wouldn't be self driving cars. It feels at once both 0:01:40.640 --> 0:01:45.280 inevitable and like hype, if that makes sense, like an artificiality. 0:01:45.760 --> 0:01:48.960 No, I mean I definitely, I definitely agree two hundred 0:01:48.960 --> 0:01:51.840 and three hundred years, that's like that's a long time, 0:01:51.920 --> 0:01:53.160 like fifty years from now. 0:01:53.280 --> 0:01:55.040 I've kind of you think fifty years from now. 0:01:55.120 --> 0:01:57.240 Well, so this is the question, which is like my 0:01:58.240 --> 0:02:00.680 is not an area I know that well, but my 0:02:00.720 --> 0:02:03.360 impression is it's like look sort of classic thing where 0:02:03.360 --> 0:02:05.720 like tech got us like ninety five percent of the 0:02:05.720 --> 0:02:08.360 way there, and then there are some edge cases that 0:02:08.400 --> 0:02:10.640 make self driving cars difficult. I don't know exactly what 0:02:10.680 --> 0:02:13.280 they are, but that getting that last five percent or 0:02:13.280 --> 0:02:16.320 whatever is so hard that it renders the whole thing 0:02:16.440 --> 0:02:19.760 very difficult, and that whatever that last percent is is 0:02:20.040 --> 0:02:23.840 the difference between the tech being like wow, versus actually 0:02:23.919 --> 0:02:24.560 changing the world. 0:02:24.720 --> 0:02:26.440 We are so close and yet so far. 0:02:26.639 --> 0:02:27.639 Yeah, it's one of those things. 0:02:27.639 --> 0:02:30.080 And I feel like again with chatbots and some of 0:02:30.120 --> 0:02:33.760 these other like current like artificial intelligence applications, it comes 0:02:33.800 --> 0:02:36.040 back to this question of like, yes, it's really great 0:02:36.080 --> 0:02:37.960 and it sort of blows your mind, but there are 0:02:37.960 --> 0:02:42.400 these hallucinations another thing, and like if it's not it's 0:02:42.400 --> 0:02:44.960 not one hundred percent reliable, does that mean it really 0:02:45.000 --> 0:02:46.680 won't be as disruptive as people expect. 0:02:46.880 --> 0:02:49.200 Well, the other thing I'm curious about is whether or 0:02:49.240 --> 0:02:52.000 not that sort of last five percent that you're describing, 0:02:52.000 --> 0:02:53.840 whether that's on the software or the hardware. 0:02:54.080 --> 0:02:56.120 Oh yeah, because that has. 0:02:55.919 --> 0:02:58.920 Implications for you know, if we do make huge leaps 0:02:58.960 --> 0:03:02.720 and artificial intell leigence, maybe that solves a software problem. 0:03:02.800 --> 0:03:06.680 But maybe the issue is actually that the sensors are 0:03:06.760 --> 0:03:10.640 too basic or too expensive, that sort of thing. 0:03:10.680 --> 0:03:12.160 I don't know the answer to any of these questions. 0:03:12.200 --> 0:03:14.080 The one other thing I'll say too, is like there 0:03:14.200 --> 0:03:15.800 is a lot of car talk these days. 0:03:15.800 --> 0:03:17.519 We've been doing more and more on the podcast. 0:03:17.840 --> 0:03:22.280 It's entirely like on the sort of EV charging side, 0:03:22.280 --> 0:03:24.959 and how are EV's and how EV production and batteries, 0:03:24.960 --> 0:03:27.320 like how are they going to reshape the industry? Or 0:03:27.400 --> 0:03:30.240 Chinese exports, how are they going to reshape the industry? 0:03:30.360 --> 0:03:32.959 It does not seem like again ten years ago the 0:03:32.960 --> 0:03:35.240 big question was like who's ahead in the self driving 0:03:35.280 --> 0:03:39.600 car race? Google, GM or Ford. There was much less 0:03:39.600 --> 0:03:41.720 talk then about EV's is the big disruptor? 0:03:42.000 --> 0:03:44.560 Right, So I think it is time for a checkup, 0:03:44.720 --> 0:03:46.640 right on what's going on with self driving cars? 0:03:46.720 --> 0:03:47.760 It is time for a checkup. 0:03:47.800 --> 0:03:50.800 And our guests says they're back, that they're happening for real. 0:03:50.880 --> 0:03:53.320 And I do believe him to some extent because I 0:03:53.400 --> 0:03:56.160 follow some people who live in San Francisco and they're 0:03:56.160 --> 0:03:57.920 tweeting about it more and more that they see them 0:03:57.960 --> 0:04:00.440 on the road. And sometimes when I'm up but four 0:04:00.440 --> 0:04:03.200 in the morning to read the Internet in the dark 0:04:03.240 --> 0:04:05.520 and drink coffee, I see like people who are still 0:04:05.520 --> 0:04:08.480 out at night in San Francisco talking about all the 0:04:08.480 --> 0:04:10.840 self driving cars around them, So there might actually be 0:04:10.920 --> 0:04:12.120 it may not be totally over. 0:04:12.240 --> 0:04:13.360 There might be they might be back. 0:04:13.840 --> 0:04:16.560 Well, the things I see on the internet about self 0:04:16.640 --> 0:04:19.240 driving cars are those edge cases where it's like a 0:04:19.320 --> 0:04:21.920 car flum mixed by a traffic cone in the middle 0:04:21.960 --> 0:04:27.200 of the street, which they're simultaneously like impressive and amusing 0:04:27.360 --> 0:04:30.120 and disappointing all at the same time. If that makes sense. 0:04:30.400 --> 0:04:31.040 It's interesting. 0:04:31.080 --> 0:04:34.359 You're very pro self driving cars. It hadn't clicked, like 0:04:34.400 --> 0:04:35.640 you really want this to happen. 0:04:35.760 --> 0:04:39.599 I have a personal self interest in not having to 0:04:39.680 --> 0:04:43.040 learn how to drive. I figure if I'm super optimistic, 0:04:43.120 --> 0:04:45.080 maybe maybe if I just hang on for like another 0:04:45.160 --> 0:04:48.400 ten years, Yeah, maybe, I don't know. Let's ask our guests. 0:04:48.440 --> 0:04:51.520 Let's ask our guests. We have the perfect guest, longtime 0:04:51.680 --> 0:04:55.640 tech journalists, a tech understander, someone who really deep delves 0:04:55.680 --> 0:04:58.880 deep into technology to understand like how things work and 0:04:58.920 --> 0:05:01.719 what's really happening. I followed his work for a long time. 0:05:01.760 --> 0:05:05.560 We're actually like we're colleagues together, like eighteen years ago, 0:05:06.000 --> 0:05:08.159 I think at a side called tech dirt. 0:05:08.400 --> 0:05:08.880 Tim Lee. 0:05:09.000 --> 0:05:13.080 He is the author of the understandingai dot org newsletter 0:05:13.279 --> 0:05:15.719 longtime tech journalist, and he recently wrote a piece the 0:05:15.760 --> 0:05:18.880 death of self driving cars is greatly exaggerated. 0:05:18.920 --> 0:05:20.720 So, uh, Tim, great to have you on the show. 0:05:21.279 --> 0:05:22.839 Hey, I'm great to be and I'm a fan of 0:05:22.839 --> 0:05:23.480 the podcast. 0:05:23.720 --> 0:05:25.279 Thank you very much, appreciate that. 0:05:25.600 --> 0:05:28.279 Let's start ten years ago, and you know, I think 0:05:28.279 --> 0:05:30.080 ten years ago there was a lot of self driving 0:05:30.160 --> 0:05:33.640 car hype and my impression was, and this is so 0:05:34.160 --> 0:05:36.440 vague and fuzzy, it's like, oh, most of it solved, 0:05:36.440 --> 0:05:37.800 but this last part's really hard. 0:05:38.640 --> 0:05:40.239 Is that true? What was that. 0:05:40.200 --> 0:05:44.719 Last part that has proven to be very challenging to 0:05:44.880 --> 0:05:47.560 like turn these from like prototypes onto track or a 0:05:47.640 --> 0:05:51.240 very like organized grid like suburb in Arizona to something 0:05:51.279 --> 0:05:52.640 that could actually be used on the road. 0:05:54.080 --> 0:05:56.200 So it is true that about ten years ago, Google 0:05:56.360 --> 0:05:59.799 was the leading company and they had vehicles that could 0:06:00.800 --> 0:06:04.000 go on certain routes with a fair amount of kind 0:06:04.000 --> 0:06:08.599 of preparation. And about six years ago Google rebranded itself 0:06:08.640 --> 0:06:11.440 as Weimo, it's self driving cart projects as Wemo, and 0:06:11.640 --> 0:06:16.240 started testing a taxi service in Phoenix, and they've been 0:06:16.320 --> 0:06:18.000 plugging away at that ever since. There were a bunch 0:06:18.040 --> 0:06:21.640 of other startups that were started between about twenty fourteen 0:06:21.839 --> 0:06:25.520 and twenty eighteen, say, and a lot of those failed 0:06:25.720 --> 0:06:28.120 or were forced to sell to some of the tech giants, 0:06:28.120 --> 0:06:31.080 And so there's many fewer companies operating in this space 0:06:31.120 --> 0:06:34.000 than the word five or six years ago. In terms 0:06:34.000 --> 0:06:36.240 of what the last little bit is, it's just a 0:06:36.279 --> 0:06:37.839 lot of little things. I mean, that's the thing about 0:06:37.880 --> 0:06:39.480 a long tail is there's a lot of stuff out 0:06:39.520 --> 0:06:42.119 in the long tail. One thing, for example that Weimo 0:06:42.200 --> 0:06:44.840 and Cruz the kind of industry leaders have been struggling 0:06:44.920 --> 0:06:47.640 with is when you deal with first responders. For example, 0:06:47.680 --> 0:06:50.400 if we hope to an active fire site, you're not 0:06:50.400 --> 0:06:53.160 supposed to dive over the hoses that firefighters are using. 0:06:53.360 --> 0:06:55.440 I mean, that's something you might only encounter every one 0:06:55.480 --> 0:06:56.800 hundred thousand miles or something. 0:06:57.040 --> 0:06:58.800 And so there's just lots of it's a really big 0:06:58.920 --> 0:06:59.840 deal to do it. 0:07:00.640 --> 0:07:02.800 Yes, absolute there's another case where a crew is vehicle 0:07:02.880 --> 0:07:05.680 like drove through a police tape in a crime scene. 0:07:05.760 --> 0:07:07.520 So there's lots of little things that I saw a headline. 0:07:07.520 --> 0:07:09.159 I haven't actually looked into this yet, but apparently a care 0:07:09.279 --> 0:07:12.240 is like drove into web codcrete. So the real world 0:07:12.320 --> 0:07:15.480 is complicated and there's just lots of weird situations that 0:07:15.480 --> 0:07:17.280 a human being because we kind of understand how the 0:07:17.280 --> 0:07:19.720 world works. You see, Oh that looks like web concrete. 0:07:19.720 --> 0:07:21.800 I shouldn'tdrive on that. But you just have to, Like 0:07:21.800 --> 0:07:23.440 it's like whack mole. You have to like hit every 0:07:23.480 --> 0:07:26.000 single like bad thing a vehicle can do it. That 0:07:26.080 --> 0:07:27.680 just takes a lot, a lot, a lot of work. 0:07:28.040 --> 0:07:29.760 Yeah, I don't know why, but I find all the 0:07:29.800 --> 0:07:34.680 stories of like robotic self driving cars behaving badly absolutely hilarious. 0:07:34.840 --> 0:07:37.200 And not the ones where they hurt people I should 0:07:37.280 --> 0:07:39.640 just caveat that, but the ones where, you know, something 0:07:39.640 --> 0:07:41.880 that we wouldn't even think about, you know, there's an 0:07:41.920 --> 0:07:44.160 object in the road, just go around it, and they 0:07:44.200 --> 0:07:46.280 seem to really struggle with I want to ask you 0:07:46.320 --> 0:07:49.040 more about why that seems to be an issue and 0:07:49.080 --> 0:07:51.320 sort of get into some of the edge cases that 0:07:51.760 --> 0:07:54.320 Joe mentioned in the intro. But before we do, why, 0:07:54.360 --> 0:07:57.400 here's a basic question. Why have a lot of these 0:07:57.480 --> 0:08:02.600 self driving car companies struggled Because on the face of it, 0:08:02.600 --> 0:08:04.680 it would appear that there is a lot of money 0:08:04.720 --> 0:08:07.720 floating out there in venture capital land that often goes 0:08:07.760 --> 0:08:12.680 into unrealistic or unprofitable unprofitable projects. So why has this 0:08:12.760 --> 0:08:15.680 been an issue for self driving cars in particular? 0:08:16.360 --> 0:08:18.600 I mean, I think on some level the basic issue 0:08:18.600 --> 0:08:21.280 with safety a lot of other areas of tech. You 0:08:21.360 --> 0:08:23.120 kind of build them in a viable product and you 0:08:23.120 --> 0:08:24.880 put it out in the world and you know it 0:08:24.880 --> 0:08:28.240 breaks sometimes, but that's fine, Like that gets you more feedback. 0:08:28.360 --> 0:08:31.640 And because you can kind of iterate rapidly, you can 0:08:31.680 --> 0:08:34.280 like scale up very quickly and get to a profitable 0:08:34.559 --> 0:08:37.720 scale pretty quickly. That obviously doesn't work if if the 0:08:37.920 --> 0:08:40.640 moving fast at breaking things is like literally breaking things 0:08:40.679 --> 0:08:43.160 and killing people. And so you have to be very 0:08:43.160 --> 0:08:45.959 close to perfect before you can launch a commercial service 0:08:46.000 --> 0:08:47.440 and start making money. And so you had a bunch 0:08:47.480 --> 0:08:49.520 of startups that were trying to do this. They had 0:08:49.559 --> 0:08:51.839 all starts of strategy to do that. Some were trying 0:08:51.880 --> 0:08:55.199 to operate in retirement communities or do like package delivery. 0:08:55.200 --> 0:08:58.840 They try to find kind of less demanding applications than 0:08:58.920 --> 0:09:01.800 like drive anywhere any time. But it's just really really hard. 0:09:01.840 --> 0:09:04.520 And so the companies that have sustained are the ones 0:09:04.559 --> 0:09:08.200 that have Amazon, Google, GM like big companies behind them 0:09:08.240 --> 0:09:09.839 who are willing to put like a billion dollars a 0:09:09.960 --> 0:09:11.800 year behind them for several years in a row while 0:09:11.800 --> 0:09:14.920 they kind of try to iron out these final little wrinkles. 0:09:16.200 --> 0:09:19.480 So zooming forward today and that makes a lot of sense. 0:09:19.520 --> 0:09:20.520 I hadn't really thought about that. 0:09:20.600 --> 0:09:22.959 It's like for many tech it's okay if there are 0:09:23.040 --> 0:09:25.640 edge cases where it doesn't work, because you just sort 0:09:25.679 --> 0:09:27.360 of like, well, you put out in the world and like, yeah, 0:09:27.360 --> 0:09:31.120 it's not a perfect product, but into minimum via, it's free. 0:09:30.920 --> 0:09:31.959 And we're refining it. 0:09:31.960 --> 0:09:33.400 It's free and we're iterating. 0:09:33.559 --> 0:09:37.280 But you cannot do that when there's big safety issues 0:09:37.480 --> 0:09:40.760 and if it's a threat to other drivers or pedestrians, 0:09:41.040 --> 0:09:44.720 it's not really an acceptable way to do product. Going 0:09:45.200 --> 0:09:48.320 to today and you are more optimistic and we'll get 0:09:48.320 --> 0:09:51.679 to that about the prospects for their existence. But has 0:09:51.720 --> 0:09:54.679 there been a breakthrough in recent years or has it 0:09:54.840 --> 0:09:58.560 just been this slow, iterative grinding away at the edge 0:09:58.600 --> 0:10:01.679 cases that it makes it so that there are fewer 0:10:01.679 --> 0:10:02.640 and fewer edge cases. 0:10:03.600 --> 0:10:06.880 I would say the second one. I mean, Weymo's technology 0:10:06.880 --> 0:10:09.800 has worked pretty well. They started doing fully driver less 0:10:09.800 --> 0:10:13.720 operations in Phoenix in the fall of twenty twenty, and 0:10:14.320 --> 0:10:18.200 I have just very gradually expanded that service now weimo 0:10:18.320 --> 0:10:20.640 I just a week or two ago they got permissioned 0:10:20.640 --> 0:10:23.520 from California regulators to begin operating commercially in San Francisco 0:10:23.600 --> 0:10:27.000 after a year or two of doing practice driving there, 0:10:27.480 --> 0:10:30.280 and so yeah, they've they've just been plugging away at it. 0:10:30.360 --> 0:10:32.559 And it's hard to tell from the outside because they're 0:10:32.600 --> 0:10:35.960 not super transparent about all the details of you know, 0:10:36.000 --> 0:10:38.120 how many incidents they have or how much work they 0:10:38.120 --> 0:10:39.360 have to do on the back end. But yeah, it 0:10:39.360 --> 0:10:41.800 seems like they're just very gradually making the technology better. 0:10:42.160 --> 0:10:44.360 And they seem to think because they're now talking about 0:10:44.559 --> 0:10:46.480 scaling up much more quickly, they seem to think that 0:10:47.000 --> 0:10:48.880 the companies seem to think that they're that this is 0:10:48.920 --> 0:10:50.440 ready for it to be a commercial product. 0:10:50.640 --> 0:10:53.120 Just a really simple question. If I were to go 0:10:53.160 --> 0:10:56.600 to San Francisco right now, could I go there and 0:10:56.640 --> 0:10:59.360 download an app or whatever and get in a self 0:10:59.440 --> 0:11:00.160 driving tax. 0:11:01.320 --> 0:11:03.480 You know, I haven't checked that recently, So it was 0:11:03.520 --> 0:11:06.040 literally like last week or the week before the California 0:11:06.120 --> 0:11:08.480 regulators gave them permission to do that, I think, And 0:11:08.600 --> 0:11:10.160 so till like last week, I think it was a 0:11:10.160 --> 0:11:12.000 waiting list. But it's definitely case, you go to certain 0:11:12.040 --> 0:11:16.200 parts of Phoenix, including the Phoenix Airport, you can hail 0:11:16.240 --> 0:11:18.520 a taxi and it's just like Uber left. 0:11:18.520 --> 0:11:19.559 You can try it. 0:11:19.640 --> 0:11:21.320 I want to do it. I want to do it, Tracy. 0:11:21.400 --> 0:11:21.680 Let's go. 0:11:22.280 --> 0:11:24.160 Let's go to Phoenix, just so that for the one 0:11:24.240 --> 0:11:25.120 ride then fly back. 0:11:25.760 --> 0:11:27.959 Sure. Yeah, I talk to people there. I mean it works, 0:11:28.080 --> 0:11:29.480 it works quite well. I mean the people. 0:11:29.559 --> 0:11:31.840 I've talked to several people who've written in those vehicles 0:11:32.000 --> 0:11:34.880 and at least in both rides they say it's flawless, 0:11:34.880 --> 0:11:36.280 that drives very comfortably. 0:11:36.440 --> 0:11:39.160 And yeah, the service they just aren't that many refuges. 0:11:39.880 --> 0:11:42.880 Can we talk a little bit more about the edge stuff, 0:11:42.880 --> 0:11:47.000 because my my impression is that, Okay, computers learn from 0:11:47.360 --> 0:11:52.160 repetition and from modeling out various scenarios, but driving is 0:11:52.280 --> 0:11:57.320 such an infinitely unpredictable experience, especially if you're in New York. 0:11:59.240 --> 0:12:01.600 You could get you get it, Like if this technology 0:12:01.600 --> 0:12:03.720 you'd never taken off, I have confidence you could do. 0:12:04.040 --> 0:12:05.880 I don't know. I think I've missed the boat on 0:12:05.880 --> 0:12:08.880 that one. But anyway, okay, but there are all these 0:12:08.880 --> 0:12:13.200 different possibilities that a self driving car could be grappling with. 0:12:13.320 --> 0:12:16.480 So for instance, an animal runs out in the middle 0:12:16.520 --> 0:12:19.880 of the street, and you know, maybe after that happens 0:12:19.920 --> 0:12:23.360 several times, the self driving car starts to realize, well, 0:12:23.400 --> 0:12:25.560 it's this animal, and then it's going to behave in 0:12:25.600 --> 0:12:27.880 this way and keep moving or stop, and I need 0:12:27.920 --> 0:12:31.080 to respond to it in a certain way. But that 0:12:31.240 --> 0:12:34.440 kind of seems to be the issue here as far 0:12:34.480 --> 0:12:35.360 as I understand it. 0:12:36.080 --> 0:12:37.760 Yes, absolutely, And there's a bunch of ways that the 0:12:37.760 --> 0:12:39.800 companies have tried to do this. So, for example, Google 0:12:39.840 --> 0:12:43.440 has long had a big test track facility out in 0:12:43.440 --> 0:12:45.439 in California about an hour I went out there a 0:12:45.440 --> 0:12:47.960 few years ago, where they have some fake roads and 0:12:48.000 --> 0:12:50.880 they'll create kind of fake scenarios. They'll have cars cutting 0:12:51.000 --> 0:12:53.360 other cars off, or have somebody like moving box of 0:12:53.400 --> 0:12:55.720 the Cross, the three People and Halloween costumes something like that, 0:12:55.920 --> 0:12:57.440 and so they try to think of what are all 0:12:57.440 --> 0:12:59.680 the situations of the self driving car could run into 0:13:00.040 --> 0:13:03.079 and kind of anticipate that. And this is also why 0:13:03.160 --> 0:13:05.600 they started in Phoenix, is one of their strategies was, Okay, 0:13:05.600 --> 0:13:07.360 there's so many ED cases, we can't do them allay 0:13:08.040 --> 0:13:10.280 all at once, and so let's start a kind of 0:13:10.320 --> 0:13:13.840 easy mode. And so Phoenix has very nice weather, nice 0:13:13.840 --> 0:13:17.679 wide streets, well marked, you know, not a lot of pedestrians, 0:13:17.679 --> 0:13:20.000 not a lot of bicyclists, and so that was kind 0:13:20.000 --> 0:13:22.560 of way most theory was that we'll do the easiest 0:13:22.559 --> 0:13:25.439 one first. The issue with that is that the economics 0:13:25.440 --> 0:13:27.160 of running a taxi service in theaters are not that 0:13:27.200 --> 0:13:30.319 great because most people already have cars, and so Cruise 0:13:30.320 --> 0:13:31.920 has kind of had the opposite approach. It's is, we 0:13:31.960 --> 0:13:34.400 want to see these educations as fast as possible, So 0:13:34.480 --> 0:13:36.800 let's start in downtown San Francisco, because that's where there's 0:13:36.800 --> 0:13:39.360 a ton of crazy situations, and so we will kind 0:13:39.360 --> 0:13:41.000 of be harder in the first place, but we gathering 0:13:41.040 --> 0:13:42.839 data very quickly and they will master it. 0:13:43.000 --> 0:13:44.040 And is not yet clear yet. 0:13:44.040 --> 0:13:46.240 I mean, both both companies now seem to think they're ready, 0:13:46.240 --> 0:13:48.280 but I don't think we've seen them in a while 0:13:48.360 --> 0:13:50.240 long enough to have a sense for kind of which 0:13:50.280 --> 0:13:51.480 of those strategies are working better. 0:13:51.520 --> 0:13:53.080 But yeah, it's really tough, and so so I. 0:13:53.080 --> 0:13:55.120 Should say, like for the first few years, both of 0:13:55.160 --> 0:13:57.800 these companies had safety divers behind the wheel of every vehicle, 0:13:58.000 --> 0:14:00.080 and so the vehicle was mostly diving itself. But if 0:14:00.080 --> 0:14:02.080 it got stuck, that says or down would have to 0:14:02.280 --> 0:14:04.840 take over. And the kind of big switching. The big 0:14:04.960 --> 0:14:07.360 risky point is when they take the first starts out 0:14:07.360 --> 0:14:09.800 of the car, which WM one did about two years ago, 0:14:09.880 --> 0:14:11.080 and cruise did I think maybe a. 0:14:11.120 --> 0:14:13.440 Year ago, and then you know, and then that is 0:14:13.520 --> 0:14:13.960 it because the. 0:14:13.960 --> 0:14:16.720 Must trickier because if the vehicle screws up, it's a 0:14:16.720 --> 0:14:17.120 big deal. 0:14:17.640 --> 0:14:19.680 I love the idea of having to train the self 0:14:19.760 --> 0:14:23.000 driving cars by like putting people in Halloween costumes in 0:14:23.040 --> 0:14:24.680 front of them, and it reminds me a lot of 0:14:24.760 --> 0:14:27.840 socializing my dog because we used to have to like 0:14:28.000 --> 0:14:31.480 wear weird hats right, or like bring balloons into the 0:14:31.520 --> 0:14:33.520 house so that he would get used to them and 0:14:33.560 --> 0:14:37.240 not freak out. But this goes back to something that 0:14:37.320 --> 0:14:40.600 I mentioned earlier, which is is the issue here the software, 0:14:40.720 --> 0:14:44.880 so like the actual modeling of the reaction to an 0:14:44.960 --> 0:14:49.920 unknown or unfamiliar event or stimulus, or is it more 0:14:50.120 --> 0:14:53.920 on the hardware side where maybe you need better sensors 0:14:54.040 --> 0:14:57.040 that are better able to appreciate the things in front 0:14:57.040 --> 0:14:57.240 of you. 0:14:58.560 --> 0:15:01.400 I would say it's more software, particularly more data. But yeah, 0:15:01.480 --> 0:15:04.040 the hardware has stayed pretty constant. I mean that the trio 0:15:04.080 --> 0:15:07.240 of sensors. Most of these vehicles have our cameras, radar, 0:15:07.280 --> 0:15:10.840 and then lighter, which is a like laser laser range 0:15:10.840 --> 0:15:12.640 finding technology that gives you kind of a three D 0:15:12.720 --> 0:15:15.920 map of your environment. And so ten years ago Google's 0:15:15.960 --> 0:15:18.080 car has had those three sensors, and I think now 0:15:18.080 --> 0:15:20.320 those sensors are better. But I don't think anybody thinks 0:15:20.120 --> 0:15:22.320 the main issues that we need to upgrade in the 0:15:22.400 --> 0:15:24.520 quality to light are really they just need they need 0:15:24.560 --> 0:15:27.640 examples of every possible educase, and they you know, it's 0:15:27.680 --> 0:15:30.160 hard to get enough of that data because some educses 0:15:30.200 --> 0:15:32.480 happen very rarely but can be very serious if you 0:15:32.560 --> 0:15:49.640 encounter them. 0:15:49.840 --> 0:15:52.520 Can you give us a quick industry overview? You know 0:15:52.600 --> 0:15:59.480 you mentioned Weimo, it's Weimo's Google Cruises, GM Cruises GM 0:15:59.640 --> 0:16:03.320 and then obviously Tesla and Elon's all Like if you 0:16:03.480 --> 0:16:05.960 just like raid Elon's Twitter feed, you would think that 0:16:06.000 --> 0:16:08.880 they've already had self driving cars like in the wild, 0:16:08.960 --> 0:16:10.920 And I don't really think that's true, but I don't 0:16:10.920 --> 0:16:12.880 really understand what's going on. Can you give like a 0:16:12.920 --> 0:16:15.640 really just sort of quick like overview who the big 0:16:15.680 --> 0:16:18.160 players are, like who owns them, and just sort of 0:16:18.200 --> 0:16:19.120 like what their status is. 0:16:19.520 --> 0:16:20.480 Yes, absolutely so. 0:16:20.720 --> 0:16:23.680 Wemo is mainly owned by Google, Cruise is mainly owned 0:16:23.680 --> 0:16:26.280 by GM. I consider Tesla to be in a different market, 0:16:26.280 --> 0:16:28.000 and some of the Tesla fans get mad at me 0:16:28.040 --> 0:16:29.880 when I say this, But Tesla is building a. 0:16:29.880 --> 0:16:31.040 Driver systems product. 0:16:31.120 --> 0:16:33.760 So all pretty much any car you drive now they 0:16:33.760 --> 0:16:36.040 have advanced cruise control where it stays in your lane 0:16:36.160 --> 0:16:37.960 and doesn't hit the car in front of you. In 0:16:38.000 --> 0:16:39.880 some ways, I think Tesla has a more advanced version 0:16:39.920 --> 0:16:41.560 of that, although also in some ways I think it's 0:16:41.600 --> 0:16:43.640 Elon must just has a low or risk tolerance and 0:16:43.680 --> 0:16:46.400 so he's kind of pushing a technology that's anyway. But 0:16:46.440 --> 0:16:48.840 so so, the key thing about the Tesla product is 0:16:48.880 --> 0:16:50.640 you were not supposed to, like crawl in the back 0:16:50.680 --> 0:16:52.520 seat and take a nap, right You're supposed to be 0:16:52.560 --> 0:16:53.840 there making sure it doesn't break. 0:16:54.240 --> 0:16:56.680 And have people crawled in the back seat and taken 0:16:56.720 --> 0:16:57.080 a nap. 0:16:57.360 --> 0:16:59.720 I'm sure somebody has that. There are videos of people 0:16:59.760 --> 0:17:02.040 doing it appropriate things well behind the wheel of the Tesla, 0:17:02.040 --> 0:17:03.760 but you're definitely not supposed to. And you know the 0:17:04.119 --> 0:17:05.480 vehicles that they have ways. 0:17:05.240 --> 0:17:07.720 Of monitoring the driver so that that doesn't happen. 0:17:07.680 --> 0:17:10.520 But anyway, So theoretically Elon Musk thinks they're going to 0:17:10.560 --> 0:17:11.600 at some point to get to the point where you 0:17:11.600 --> 0:17:12.800 don't have to be behind the wheel. But I do 0:17:12.880 --> 0:17:15.240 not think they're close to that or really laying groundwork, 0:17:15.280 --> 0:17:17.320 because one of the things for any service like that 0:17:17.400 --> 0:17:19.680 is you need an operation staff because the vehicle is 0:17:19.760 --> 0:17:21.919 occasionally get to get stuck, and when that happens, it 0:17:21.960 --> 0:17:23.280 used to be able to phone home and get kind 0:17:23.280 --> 0:17:25.240 of remote guidance about how to deal with it. And 0:17:25.280 --> 0:17:26.879 as far as they know, Tesla's not doing it anyway. 0:17:26.880 --> 0:17:29.800 So that's Tesla and then the other two companies. There 0:17:29.800 --> 0:17:31.640 are a few other companies I would say a little behind. 0:17:31.640 --> 0:17:33.600 So Amazon has a company called Zookes that used to 0:17:33.640 --> 0:17:35.560 be a startup but acquired by Amazon a couple of 0:17:35.600 --> 0:17:36.040 years ago. 0:17:36.520 --> 0:17:38.360