If you have the chance to try out a Waymo (you own a credit card and smart phone and find yourself in San Francisco, Los Angeles, Phoenix or Austin if you can get through that waitlist) I thoroughly recommend it.
Right now it's the most exciting tourist attraction in San Francisco.
The first few minutes are pretty terrifying... but the ride is so smooth that you very quickly settle into it. It's absolutely worth experiencing.
I was in SF back in May, but didn't manage to get through the waitlist :(
It was so cool to see them diving around.
Interestingly, when I showed the clips to some of my senior family members, they didn't seem interested at all. I think they couldn't comprehend what was going on, even after I explained.
Their (several independent trials) reaction was similar to showing them some AI-generated image of something which clearly can't exist. It was so absurd that it was just filtered out with a comment "yeah, yeah - nice car".
I'd like to but how do you overcome the slight powerless feeling? Like when those guys surrounded a vehicle and made it to stop with people helplessly inside. You're kind of stuck.
You can open the door and get out if it's stopped. You can press the "pull over" button to have it pull over as soon as it can safely do so. You can call rider support or 911 while the doors remain locked to the outside.
I agree that it's a slightly more vulnerable feeling than driving myself, but I still feel far less vulnerable than taking public transit or walking. And when you factor in risk of a crash, it's also safer than Uber or driving myself.
I don't think Waymos will often be targeted by criminals for anything besides vandalism, with 20+ cameras and potential remote monitors the risk:reward ratio is much worse than other options.
I did my first ride when I was at a conference in SFO earlier this month. Part of it was it 'shows' you what it sees, which was a bit better on what I could track as a human with my head on a swivel. The car never went beyond residential speeds and it was very much felt like a fast car trying to be cautious. It was less terrifying then teaching my kid to drive. We very much felt like we were on a tourist ride - and the taxis that drove past us often had folks trying to get video/pictures of us passengers in a car without the driver.
Watching a couple hundred of them try to get out of a parking lot was comical.
The real difference is that those guys felt emboldened to act that way with nobody at the wheel. They might not have trusted an uber driver to not drive into them, and there is a social element where some people apparently don’t think about how they are acting on video, they think nobody is around.
I expect this to go away over time as people realize driverless =\= witnessless
You can quickly call for human assistance on one of the big support buttons.
You get out and use reasonable force to break free of their false imprisonment, the same way you would if you were in your own car and you were falsely imprisoned.
I hope very strenuously that the implication is not that you would run over people who are blocking you if you were in control of the car.
This is certainly true, but doesn't mean that they aren't hugely popular [1]. We (me with young children) used to call out "Driverless Car" whenever we'd see one, but it is less interesting when you may see 15 over the course of a 2.5 mile, 15 minute drive.
I certainly understand the sentiment, but can you think of historical examples where technology, which oftentimes in history has undermined income opportunities for people, has not been adopted for that reason? I guess I'm really asking whether there has ever been a successful effort to stop the adoption of tech that undermines human labor/income opportunity.
We're certainly in the thick of it right now. Recent high profile ones: LLMs, port automation, other types of AI for media production, I'm sure the list goes on and on.
The reason I'm asking is not to advocate for technology, but rather to understand if there have been successful efforts in the past to stifle tech. If there haven't been noteworthy ones, then my gut says we ought to be focused on how to get those people to not be left behind (so so incredibly difficult) rather than preserving something solely for the sake of people making money off of it. Like I said, so easy to say, but incredibly difficult to achieve in practice. Not sure there are great examples of people not being left behind.
This is really a valid concern. We talk about stable wages for Uber and meal delivery drivers here in Seattle, and when business goes down as a result of policies that pay drivers more but cost users more, we say that's just a society cost (don't use the service if you can't pay the 20% tip!). But if Waymo ever moved in here, I'm sure there would be a bunch of people on the far left crying bloody murder since it upends their whole feedback loop.
That's an interesting tradeoff to make. So try it once and swear it off afterwards for ideological purposes? I already feel myself creating a bit of a conversational rage trap here though, so apologies in advance.
I have mixed feelings on that video. The general bit-piece feel on self driving is not really deserved, IMO, and seems to be against self driving for the sake of it.
I think self driving will be a massive, massive boost to human productivity as it gets better, but we shouldn't use it as an excuse not to continue pursuing other means of transport. It doesn't really fix any of the scaling problems with car transport regardless.
> Right now it's the most exciting tourist attraction in San Francisco.
I'm as excited as you about self driving cars, but the vehicles are essentially driven remotely, maybe not by live video feed but nonetheless by remote operators, in the sense that matters. It is more tourist attraction than it is end-all be-all of technology, which was very deflating to find out.
All I am really saying Simon is that you are a highly educated guy: adopt a more nuanced take on this. According to the people I know who work in this space and are not trying to raise money for autonomous vehicles, and according to many journalists, there is a lot of consensus that Waymo has adopted a remote driving scheme that is working, and perhaps that is why they are operating a taxi service and others are not. It isn't clear if that growth story, as great as it is, will help them raise enough money to invent truly autonomous vehicles.
Vehicles per-operator is the important metric, imho. If one operator can supervise a dozen vehicles at a time with only occasional manual intervention then it's still a massive labor-saving tool.
The quality of app drivers in LA is wanting. I’ve been proselytized, informed of imminent apocalyptic civil wars, asked if I know any guys who want to buy copper in bulk. And how can I forget the driver asleep at the wheel of a Tesla with regenerative braking on max, with each doze into slumberland jolting the whole carriage into alert with a smart ping of the battery charging, all the long way from downtown to the airport? I don’t care if it’s double the price and I occasionally get stuck behind an errant traffic cone for an hour, I will gladly take Waymo.
Assuming increasing number of waymo cars for the next four years, I wonder how this is going to affect the so called "Car-Free LA" goal for the 2028 olympics. Maybe it'll just be a tweak in a marketing of what it means to be "car free."
I doubt there's ever going to be a big bang moment where self driving just works everywhere. Instead it's going to be a slow rollout of self driving taxi's in increasingly different environments.
Of course. We're still there with the horses -> cars transition. There are still plenty of places that horses can get to but cars can't.
And similarly: the transition will also go in the other direction. We'll start making roads and navigation easier for self-driving cars and prioritize the destinations we care about the most. At the tail end of the transition, there will still be areas with ridiculous intersections and confusing rules that only humans can do. We just won't care about them as much by then because everything we do care about is reachable by self-driving cars.
I think it's actually going to be the other way around. We'll build infrastructure exclusive to AVs where the rules are too complicated for humans, but allow AV traffic to move more efficiently. For example, an AV shouldn't need to stop and wait at a red light if there's no traffic to wait for.
Roads are already easy to drive. We already have signs on roads directing people to the most prioritized destinations. And now we have navigation apps that can even provide turn-by-turn and lane-by-lane directions. The issue is that most people simply ignore traffic rules when it is convenient to do so.
We didn't redesign roads and navigation for taxis or rideshares, and we're not going to do that for self-driving cars either.
At 2 trips per day for 300M Americans over 7 days, that would put the rideshare takeover at ~4.2Bn. If we extrapolated based on the referenced graph and exponential growth, that would put the takeover at 2029 :)
Its safe to assume that the limiting factors will soon become sourcing of components of the perception and control stacks.
I'm not going to get excited until this can drive in Boston/Cambridge. If it can navigate that nightmare (and not kill any cyclists) I'll be impressed with what self-driving cars can do. Bonus points if they can work in snow when sightlines are obscured.
They have been navigating the SF Tenderloin for over a year. Entire streets are a crosswalk. Most biologic drivers would barely function there. But I do wonder how lidar/visual integration works in snow.
Absolutely. Every country, state, city, county can enact their own regulations as they see fit. This is Uber rollout on steroids, imo. There are some many hurdles to making this happen even if the tech exists.
In places that have real winters, there are going to be significant blocks of time where self-driving will never work unless real synthetic intelligence is reached.
Way too much of this is developed in places that... don't have weather.
When you stomp on the brakes in a reasonably new car that's skidding on ice and it makes a loud thunking noise, what is that, and what's controlling that system?
AV companies test in winter conditions. Waymo had NYC and Buffalo deployments years ago, and they do winter testing in Tahoe and Michigan. This winter they'll do the upper peninsula, Detroit and Tahoe again.
I'm not sure what you consider "real" weather, but those are all places I think qualify by any reasonable definition.
I can't suppress the strong feeling that they will, not by being technologically mature, but by easing a lot of regulations that are restraining them now.
Especially with Tesla's Ceo soon to be a member of the government.
Just like cars companies made jaywalking illegal or bought public transport companies to close them.
Is Tesla a serious competitor in this space? I know they've made a lot of noise about self-driving cars, but that all seems to be hype generated to sell more human-driven cars. Maybe I'm off base here. I don't really follow Tesla closely.
It still won't be a big bang moment. There might be a moment you can go hands off on highways, and a moment you can go hands off in urban environments. But it's still going to be gradual overall.
Tesla FSD already works everywhere, even on unpaved roads. It just doesn’t work as well as Waymo.
Waymo works very well, just not in as many places as Tesla.
You might bet on Waymo because they have a fully working product already, but I’m betting on Tesla because of the vast amount of training data they are collecting. There’s a bitter lesson here.
> the vast amount of training data they are collecting
They keep pushing this point. And they do appear to be collecting an absolute firehose of data from the millions of vehicles they have on the road. By comparison, Waymo collects a lot less data from many fewer vehicles.
Which leads to some tough questions about Tesla's tech. If they have (conservatively) 10x the training data that Waymo has, why can't their product perform as well as Waymo? Do they need 100x? 1,000x? 10,000x?
Assuming they were at parity with Waymo today, this would suggest that their AI is only at best 10% as effective as Waymo's, and possibly more like 1% or 0.1% or whatever. But since they can't achieve parity, it's not even possible to bound it.
It's entirely possible that their current stack cannot solve the problem of autonomous driving any more than the expert systems of the 60s could do speech translation.
I haven't heard a compelling argument as to why a system that is at best 10% as effective would ever be expected to be the leader.
FSD and Waymo are completely different products. FSD isn't even autonomous, as the user manual reminds you:
Always remember that Full Self-Driving (Supervised) (also known as Autosteer on City Streets) does not make Model Y autonomous and requires a fully attentive driver who is ready to take immediate action at all times.
I hope for the best for Tesla, but they are many years behind Waymo. The world definitely needs a second working self-driving system! Right now comparing Tesla and Waymo is nonsensical. Once you can sit in the backseat of a Tesla while it drives there might be some worthy comparisons to be made.
My definition of "works" includes the fact that a self-driving car will never drive into a parked fire truck, or many other things i've seen tesla FSD do.
I don't understand the "bitter lesson" reference here. The bitter lesson is that general methods of computation are more effective. How is one of the two not using general methods?
One challenge for Tesla and Waymo has been the piecemeal permitting process. Even though California gave Waymo a statewide permit they have still needed to work through various cities/counties for permits. I imagine one goal of Musk's is to make that all go away sometime next year. I'm not making a comment on whether I agree that is a good idea. Just speculating.
If training data is such an edge for Tesla, how is it that Waymo works so much better than FSD with only 1/1000th the data?
I also don't see any evidence that Waymo can't work anywhere. They recently expanded to Austin, and it seems that it immediately drives better than FSD.
Tesla FSD will never catch up to Waymo until they switch to LIDAR and have human assistance when the vehicle gets into complicated scenarios such as emergency vehicles blocking the road and redirecting traffic.
I'm betting on Tesla not for the technology, but because President Quid Pro Bro is probably going to issue an Executive Order that turns every Tesla company into a federally blessed monopoly.
Training data is trivial to collect. Betting on Tesla because they have more training data than Waymo is like betting on Roscosmos because they have more employees than SpaceX.
Waymo was able to do this with less miles. How much data does Tesla really need at this point? Assume you have all the data in the world that you could ever possibly want. How much of that can you really compress into a car for real FSD?
Waymo vs Tesla definitely smells like bitter lesson to me yes, 100%. With Waymo being on the bitter side, to be clear. Future will tell if the intuition is right on this one
Looks like Waymo will be the first Alphabet moonshot that will reach 100B valuation in the not too distant future. Just need to hit the most profitable cities and Uber will have to compete at lower prices, while they're a public company. The nest egg will will be kept full as it's the only working autonomous ride solution out there making any money.
I think it's great we finally have self-driving cars, but why the huge valuation? Waymo has the enormous capital cost of owning the cars. Unlike taxi companies they don't push that capital cost to their drivers.
I think it's great we finally have self-driving cars, but why the huge valuation? Waymo has the enormous capital cost of owning the cars. Unlike taxi companies they don't push that capital cost to their drivers.
Ultimately, it's still the riders who pay for the drivers and their cars in a traditional taxi company and Uber. The cost of the car doesn't just disappear.
> Well currently Waymo is about double the price of Uber.
This is not accurate in my experience in SF. Most of the time, the price is comparable to UberX and during peak-times, the price is upto 20% higher than UberX.
Right this second, 11 am Tuesday after Veterans day, 2024, a trip to the Palace of Fine Arts would cost me $15 on Waymo, $24 for a wait and save on Lyft, and $17 for an UberX.
I also hope it will be cheaper, but it's definitely worth paying more for Waymo because of the better experience: safer, smoother, quieter, no weird smells, or conversation, etc.
The losses are currently in the neighborhood of $1B per year. That may or may not reduce... scaling is hard. For every car they add right now, the worse the loss becomes. So I'm a little skeptical on your valuation estimate.
WARNING: never try to take Waymo to the Hollywood Bowl.
Their app lets you book a ride to the entrance... then stops and makes you exiut, in front of a motel, blocking its driveway, a mile from the venue.
There is an invisible "exclusion zone" there that is not marked on maps, and not communicated to you in any way. You just get dumped somewhere other than the displayed destination and there is no ability to "add a stop" or otherwise modify the ride.
Ask me how I know when trying to make a show... or ask the people in the other Waymos we watched pull up as were frantically trying to get a Lyft.
Not sure how long ago you did this, but just opened the app and tried to book one there and it says I can't because it's outside the service area (it shows Hollywood Bowl but in the greyed out "Results outside of service area."
I've taken a Waymo to several large events recently and usually for those it tells you when booking how far from your destination you're actually going to be dropped off or picked up. Maybe a bunch of those folks, including yourself, complained and they fixed this issue for the Bowl.
It's long, but well worth an hour of your time. He even states in the comments that he set out two years ago to make a positive video about AVs:
What was most surprising to me was that when I began researching this video (two years ago!) it was going to be about some of the technical challenges that would need to be overcome in order to make self-driving cars a reality, but the conclusion was going to be that ultimately, AVs would be a good thing.
By the time I was done researching this topic I was absolutely horrified of our future self-driving dystopia.
The future he illustrates seems mostly plausible, except that it depends on all of the technology functioning flawlessly. I have a hard time believing that streets full of high-speed AVs functioning in perfect synchrony is likely.
However, that doesn't change my general agreement with the conclusions he draws in that video and the rest of his channel.
LA and SF in no way encapsulate the whole of global cities. No snow, easily navigable grid systems, wide American lanes, English road signs, best google maps coverage, etc. Once they are operating in Rome, Lahore, Seoul, etc. then I’d be more convinced.
Can't wait for them to move up to Seattle. We're basically San Francisco with slightly easier hills and more rain/occasional ice, so seems like a good candidate.
Google and Waymo have been working on that problem for a while. And starting a Waymo service area involves letting loose a fleet of them there. You should expect the map quality to improve fast in serviced areas.
The Chinese techs have already launched self driving taxis in China. And China traffic can be really chaotic, although I'm not sure if they aren't launched in any old urban areas yet (e.g. not new towns with better roads and less chaotic traffic).
Readit News
Right now it's the most exciting tourist attraction in San Francisco.
The first few minutes are pretty terrifying... but the ride is so smooth that you very quickly settle into it. It's absolutely worth experiencing.
It was so cool to see them diving around.
Interestingly, when I showed the clips to some of my senior family members, they didn't seem interested at all. I think they couldn't comprehend what was going on, even after I explained.
Their (several independent trials) reaction was similar to showing them some AI-generated image of something which clearly can't exist. It was so absurd that it was just filtered out with a comment "yeah, yeah - nice car".
I agree that it's a slightly more vulnerable feeling than driving myself, but I still feel far less vulnerable than taking public transit or walking. And when you factor in risk of a crash, it's also safer than Uber or driving myself.
I don't think Waymos will often be targeted by criminals for anything besides vandalism, with 20+ cameras and potential remote monitors the risk:reward ratio is much worse than other options.
Watching a couple hundred of them try to get out of a parking lot was comical.
I expect this to go away over time as people realize driverless =\= witnessless
You can quickly call for human assistance on one of the big support buttons.
Is that a common thing? I wouldn't be surprised honestly. I heard of the traffic cone thing but not this.
I hope very strenuously that the implication is not that you would run over people who are blocking you if you were in control of the car.
Deleted Comment
Right now those robocars are widely and openly hated by many people who live in San Francisco
[1] https://www.reddit.com/r/SelfDrivingCars/comments/1gi8cyf/wa...
I bet if you did polling the vast majority of people would either like them or be ambivalent.
We're certainly in the thick of it right now. Recent high profile ones: LLMs, port automation, other types of AI for media production, I'm sure the list goes on and on.
The reason I'm asking is not to advocate for technology, but rather to understand if there have been successful efforts in the past to stifle tech. If there haven't been noteworthy ones, then my gut says we ought to be focused on how to get those people to not be left behind (so so incredibly difficult) rather than preserving something solely for the sake of people making money off of it. Like I said, so easy to say, but incredibly difficult to achieve in practice. Not sure there are great examples of people not being left behind.
Don't worry. Once the novelty has worn off, the cost will probably be on par with Uber and Lyft with real human drivers --- if not higher.
The acquisition, maintenance and supervision costs for the self driving vehicles will probably consume any difference in cost.
I think self driving will be a massive, massive boost to human productivity as it gets better, but we shouldn't use it as an excuse not to continue pursuing other means of transport. It doesn't really fix any of the scaling problems with car transport regardless.
>Some say that we have imprisoned thousands, but the reality is that we have freed millions.
Waymo is very much like this. It inconveniences a few people, but it also saves a lot more from accidents, or even worse.
I'm as excited as you about self driving cars, but the vehicles are essentially driven remotely, maybe not by live video feed but nonetheless by remote operators, in the sense that matters. It is more tourist attraction than it is end-all be-all of technology, which was very deflating to find out.
All I am really saying Simon is that you are a highly educated guy: adopt a more nuanced take on this. According to the people I know who work in this space and are not trying to raise money for autonomous vehicles, and according to many journalists, there is a lot of consensus that Waymo has adopted a remote driving scheme that is working, and perhaps that is why they are operating a taxi service and others are not. It isn't clear if that growth story, as great as it is, will help them raise enough money to invent truly autonomous vehicles.
Dead Comment
> Los Angeles County there are 7.67M registered automobiles and trucks
what are you even talking about
And similarly: the transition will also go in the other direction. We'll start making roads and navigation easier for self-driving cars and prioritize the destinations we care about the most. At the tail end of the transition, there will still be areas with ridiculous intersections and confusing rules that only humans can do. We just won't care about them as much by then because everything we do care about is reachable by self-driving cars.
LLMs are already better at this than humans:
https://x.com/petergyang/status/1707169696049668472
I think it's actually going to be the other way around. We'll build infrastructure exclusive to AVs where the rules are too complicated for humans, but allow AV traffic to move more efficiently. For example, an AV shouldn't need to stop and wait at a red light if there's no traffic to wait for.
We didn't redesign roads and navigation for taxis or rideshares, and we're not going to do that for self-driving cars either.
At 2 trips per day for 300M Americans over 7 days, that would put the rideshare takeover at ~4.2Bn. If we extrapolated based on the referenced graph and exponential growth, that would put the takeover at 2029 :)
Its safe to assume that the limiting factors will soon become sourcing of components of the perception and control stacks.
Way too much of this is developed in places that... don't have weather.
I'm not sure what you consider "real" weather, but those are all places I think qualify by any reasonable definition.
Especially with Tesla's Ceo soon to be a member of the government.
Just like cars companies made jaywalking illegal or bought public transport companies to close them.
Tesla FSD already works everywhere, even on unpaved roads. It just doesn’t work as well as Waymo.
Waymo works very well, just not in as many places as Tesla.
You might bet on Waymo because they have a fully working product already, but I’m betting on Tesla because of the vast amount of training data they are collecting. There’s a bitter lesson here.
They keep pushing this point. And they do appear to be collecting an absolute firehose of data from the millions of vehicles they have on the road. By comparison, Waymo collects a lot less data from many fewer vehicles.
Which leads to some tough questions about Tesla's tech. If they have (conservatively) 10x the training data that Waymo has, why can't their product perform as well as Waymo? Do they need 100x? 1,000x? 10,000x?
Assuming they were at parity with Waymo today, this would suggest that their AI is only at best 10% as effective as Waymo's, and possibly more like 1% or 0.1% or whatever. But since they can't achieve parity, it's not even possible to bound it.
It's entirely possible that their current stack cannot solve the problem of autonomous driving any more than the expert systems of the 60s could do speech translation.
I haven't heard a compelling argument as to why a system that is at best 10% as effective would ever be expected to be the leader.
I hope for the best for Tesla, but they are many years behind Waymo. The world definitely needs a second working self-driving system! Right now comparing Tesla and Waymo is nonsensical. Once you can sit in the backseat of a Tesla while it drives there might be some worthy comparisons to be made.
Tesla works nowhere as a fully autonomous vehicle.
I also don't see any evidence that Waymo can't work anywhere. They recently expanded to Austin, and it seems that it immediately drives better than FSD.
I would not bet on Tesla's FSD other than on highways. Same as many of the Tesla FSD owners I know.
Deleted Comment
I'll take that bet...
I predict the Chinese, in a decade, will have the first FSD
Tesla is miles behind
Plus although they have the capital cost of owning the cars they don't have the COGS of paying the drivers so presumably much higher margin per ride.
Uber without the expense or problem of a driver.
This is not accurate in my experience in SF. Most of the time, the price is comparable to UberX and during peak-times, the price is upto 20% higher than UberX.
I also hope it will be cheaper, but it's definitely worth paying more for Waymo because of the better experience: safer, smoother, quieter, no weird smells, or conversation, etc.
Their app lets you book a ride to the entrance... then stops and makes you exiut, in front of a motel, blocking its driveway, a mile from the venue.
There is an invisible "exclusion zone" there that is not marked on maps, and not communicated to you in any way. You just get dumped somewhere other than the displayed destination and there is no ability to "add a stop" or otherwise modify the ride.
Ask me how I know when trying to make a show... or ask the people in the other Waymos we watched pull up as were frantically trying to get a Lyft.
I've taken a Waymo to several large events recently and usually for those it tells you when booking how far from your destination you're actually going to be dropped off or picked up. Maybe a bunch of those folks, including yourself, complained and they fixed this issue for the Bowl.
https://www.youtube.com/watch?v=040ejWnFkj0
The future he illustrates seems mostly plausible, except that it depends on all of the technology functioning flawlessly. I have a hard time believing that streets full of high-speed AVs functioning in perfect synchrony is likely.
However, that doesn't change my general agreement with the conclusions he draws in that video and the rest of his channel.
It's not only because of the chaos but also that maps aren't accurate enough.