Human drivers are to blame for most serious Waymo collisions
understandingai.org
understandingai.org
> I assume that safety will only increase as the proportion rises.
I assume that safety will drastically decrease as self-driving cars are introduced.
I live at a train stop in SF. I take a bus to work. Public transit is great. I still use Waymo sometimes because coverage isn’t always amazing at all times. Sometimes I’m in a rush, sometimes I want some privacy, sometimes I’m carrying some heavy things.
Replacing low-occupancy human driven vehicles with low-occupancy robot driven vehicles is good if they’re safer. Today, they’ve proven to be safer. Today, lives are saved because a human driver chose not to drive. If a private individual wants to pay more (<2x, not a single order of magnitude), then let them. There is no reason that the private introduction of self driving cars needs to stop the investment in public transit.
Let us save lives on the road to the future, even if some stops along the way aren’t your chosen stops.
I haven't seen that data. Do these cars drive in rainy, wet, snowy weather or in cities with crazy streets such as the northeastern United States? Not that I know of. I think it's a wild leap to suggest they're safer already today.
What is so magical about L5 criteria? Today Waymo is performing drives on public roads, and it’s safer than a human. Today. With whatever Lx criteria it meets. The label is irrelevant, all that matters is the mile-per-mile record of being safer.
Not sure what “Waymos own admission” means, but they out perform not underperform.
https://waymo.com/blog/2023/12/waymo-significantly-outperfor...
Parent's point is that in the context where they do work, they save lives. You're arguing about something else entirely - whether they work in all situations. But that's irrelevant.
As a fellow driver, I don't care that they can't drive out of that area, I care if tf they're a safer driver.
As such, L5 seems to me like the wrong bar.
And, of course, if you otherwise basically never drive, you're probably a hazard in those marginal conditions.
They systematically get in less accidents, the accidents they do get in are not their fault, and they tend to be less dangerous when they occur. As I understand it, that unambiguously means outperforming.
We don’t need to remove every manual car in the world overnight, but this seems like a clear win, at least in the regions they operate. They don’t need to cure cancer and drive in every road condition at all times in all places and also not ever make any mistakes at all to be an improvement. “L5” is just a label, which I guess doesn’t apply, while “safer”, “less accident prone” are different labels, which do apply.
Nor does autonomy have to be limited to car-sized vehicles. It could be used for larger vehicles, perhaps intermediate between car and bus size — once you don't have to pay for a driver, operating a larger number of smaller vehicles is much more viable. These could be dynamically dispatched and routed, eliminating many drawbacks of existing bus service.
However, for your "door to door" point, that's not desirable anyway. Even here in Tokyo, you have to walk a lot to get between your home and work or wherever you're going. That's a big, big part of why people here are so much thinner than Americans. Having transport that takes you door-to-door without any significant walking is terrible for your health.
i wonder how much of a different technical problem that is from what they do today...
Surveillance state would love that! Hundreds of thousands of cars with 360 degree cameras and sensors capturing everything inside and outside the vehicle. Forget installing CC cameras. just claim “nAtIoNaL sEcUrItY”, get a rubber stamped warrant from FISA court, and go to town.
A problem created by cars (pedestrian vs car, MVA accidents) may be solved by self driving cars. But we introduced a perfect system to allow a foreign or domestic state to perform over reaching surveillance. Ripe for abuse as well.
All of this nonsense could have been avoided if cities were just built with alternative transportation methods in mind. But nope, we have to use cars for every stupid little activity.
Wasteful. Polluting.
1. If you judge new technology by the absolute worst way that it can be used, you are going to spend your life worrying about a million things that will never happen. Focus on stuff that's happening instead of hypotheticals.
2. In a really dystopian setup, you can be restricted from leaving "your zone" regardless of autonomous vehicles. There's plenty of existing technology that can help with that. (number plates, cellular networks, facial recognition, etc.)
Like cameras and AI used to enforce the “not allowed to sing while driving” policy for Amazon drivers?
The only way to make sure things never happen is by worrying about the worst possible misuse of the technology. Because someone will misuse it that way if it gets him money or power.
I'm not saying to be naive to possibilities, especially as they begin to move from the "possibilities" bucket to the "realities" bucket, but isn't it better to focus on making progress against stuff that's already happening? If Amazon are preventing drivers from singing, I think those drivers might be able to change that.
How do you prevent "hypotheticals" from becoming "stuff that's happening" if you never pay any attention to them in the first place and let them happen?
Once it's starts happening, speak; if speaking doesn't work, fight; if fighting doesn't work, move. This works.
As a computer, i find your trust in technology amusing. /s
Comparing to all drivers is one thing, but comparing to ride shares is another and yeah, Uber and Lyft are done.
The market is definitely going to choose Waymo.
The new reservations of a driverless car arent going to be big enough to trump the distrust and negative experience of other humans, given the option. The pricing is comparable, and the rest of the experience is better. The driverless part also seems trained on the driving customs of the area, here in LA it was assertive when it needed to be and overall impressive. LA has some sketchy left turn customs and the car did it fine.
As I understand it, a lot of women in the areas where Waymo's currently deployed feel safer without having a driver as well.
Until criminals figure out they're sitting ducks.
Given the tendency for premature HN opinions that are absolute to often swing the other way in practice, I'll be curious to see how this actually goes.
They’re doing all the right things, selective city rollout, FOMO inducing waitlist once there, offering great service once in
Another important thing to think about is that Uber and Lyft are operated by people. Often these working people need the meager salary to live - wouldn't you rather support your local economy instead of being a research subject?
Any business that relies on charity over market validation is done.
Waymos don't even ask for tips, the debate about it has been completely leapfrogged.
I spent the last 10 years riding ubers, the next 10 in waymos. Very easy call.
Uber and Lyft subsidized their service with VC capital for a long time as well
Do you do the same in LA for left turns, and did Waymo find a replacing clue?
It’s kind of sketchy and probably legally dubious, but it’s the only way to turn left on busy intersections without a protected left. I’d feared that a Waymo would get stuck in such situations indefinitely, but they seem to be perfectly comfortable performing these left turns as a human would.
then it wont be legally dubious
I’d consider my dad a good driver - he’s driven many hundreds of thousands of miles without a crash in probably 30 years. Does such data exist?
https://waymo.com/safety/impact/#downloads
There isn't much to the data available for download, but it looks like 0.00001207261588 accidents per mile, or ~1.2 accidents per 100,000 miles (268/22199000). Figuring your father drives 15k miles per year, times 30 years and rounding up to 500k miles, Waymo has a recorded 6 accidents to your father's 0.
Not sure why that's an interesting comparison, however.
Assuming your dad is good at not driving when he shouldn't (tired/drunk/angry), he's not on the road when it's worrisome. I don't worry about getting into accidents with drivers who aren't on the road, I worry about the tired/drunk/angry drivers I do have to share the road with. Waymo at 2:15am after the bars let out is much less worrisome than any other car at that time, because I have no idea who's in that other car. Your father could be the safest driver ever, but I have no idea if it's him in the other car, or if that driver is totally blacked out and shouldn't be driving.
I think it’s interesting because:
1) it gives Waymo a higher target to shoot for - it hasn’t “solved” self-driving because its safer than the average driver. I am so impressed by Waymo, but I feel like some of this article smacked of premature “mission accomplished” vibes. The fact that it just accepted the comparison to average without caveat is an example of that. 2) As a matter of policy, everyone can agree that a Waymo ride home for the tipsy is good, but where policy will have issues is convincing good drivers such as my dad to take Waymos everywhere. Not to mention most drivers irrationally think they’re way better than average - that will affect policy in a real way.
I wouldn’t be surprised if a top x% of drivers would outperform Waymo at this current juncture. (Especially perhaps in things like heavy rain).
However, I’ll use myself as an example. I’ve driven over 1.2 million miles over the last 17 years. (It may be closer to 1.5 million but I’m only counting miles on vehicles I’ve owned as that’s easier for me to calculate quickly ). Without an accident.
I know that I get tired and my driving skills drop[1]. I have to sneeze or cough and they drop. I’m on a long continuously straight road, stressed about xyz and they drop.
Self driving cars won’t necessarily have the same short comings.
Even the self driving features that my car has (very limited, 2023 Infiniti Q50 Red Sport 400) are better at times then I am. (Though it often will wait far too long to brake in my opinion when using the auto cruise even at maximum distance).
However, I do think humans have advantages too in some situations. (If you can on average track what’s around you and try to think as close as possible to 12 seconds ahead…. If you haven’t try it sometime on the highway. )
[1] perhaps luckily, I have a tendency to drive slower when fatigued. However at times I’ve pushed it, realized I’m going 15 mph below the speed limit, and realized I needed to not be driving. I also know that when not fatigued my lane assist will usually trigger 0 times on a normal 20-30 mile drive. When fatigued, it may begin to trigger once every 10 minutes. To clarify, I’m not out of my lane, but I’m moving out of center.
How the top 10% of drivers are going in amongst all that isn't really a factor as far as I can see. They'll probably end up banned from taking the wheel at some point for consistencies sake but they are ultimately not really a factor. Besides, automated cars will overtake (hehe) their skills at some point whether we track it specifically or not; in the long term humans can't compete against an engineered process.
In conclusion, Waymo, being a taxi service, should compare their safety record to that of human taxi drivers.
Also, a taxi driver may not be a good baseline since what you're replacing is not exactly taxi drivers but instead you're replacing some taxi drivers but also yourself.
(2) Uber drivers have a reputation for being extremely aggressive and don’t need a special license. I’d also like to see how they compare.
(3) I do want to see the data you propose, because I think you’re fundamentally right.
For some reason, not having solved every possible angle of driving from day 1 makes it seem like the technology isn't worth pursuing.
But let me ask this: do you have ABS on your car, or do you pump the brakes to stop when it's bad out like it's still 1980?
This is because nhtsa's data comes from accident reports and insurance claims, and people get into fender benders everyday that never get reported because nobody wants their rates to go up.
I remember reading that these Waymo cars haven't been allowed onto highways until now. If thats true I wonder how that will change these statistics since that dramatically increases the need for the vehicle to think and perform...
Living in Silicon Valley I've been passing those cars for a decade without the opportunity to ride. Now they even have regulatory approval for this area, but still no service and no timeline for service.
If you look at some of the skepticism around self-driving cars in the US, it makes sense that they're going slow and careful.
Even if a spike in accidents was still lower than the equivalent for humans, that would bring a lot of political scrutiny to Waymo that would then slow their deployment.
It reminds me of the kids that die due to their parents fear of vaccines
The stakes are very high and public tolerance for error is basically zero.
If instead it had turned out that their system was well built but ran into a freak edge case, I doubt they would have shut down.
Sam Peltzman showed with statistics that the FDA's mandate of "safe and effective" slowed down the development of new pharmaceuticals past the point that there were more deaths.
He took it to Congress, who simply latched on to "safe and effective" as an obviously good idea, statistics be damned.
"Regulation of Pharmaceutical Innovation", Sam Peltzman https://www.amazon.com/Regulation-Pharmaceutical-Innovation-...
The next step is to start building models of human minds (as is modeling human behavior, not literal mind uploads), as well as models of every self driving car model, and do full on global (-ish, limited to the general area) optimization of outcomes according to a publicly available and audited decision theory and utility function.
It's the only way to enable superhuman avoidance actions without making things worse by confusing others. This is why the models of humans and other robotic cars are needed.
But clearly it is not publicly available or audited.