It's very interesting, because on one side we have clear data showing that autonomous cars at the Waymo level are safer than humans.
On the other hand we don't give technology the same pass we give to humans, meaning that we don't tolerate machine incidents the same way we tolerate human ones.
Which means that self-driving cars need to be a magnitude of order safer than humans.
And even though a self-driving car will never road rage or use the car as a weapon on christmas markets, we'll still tolerate humans driving more than one odd 1-in-a-billion ai-error caused incident.
> or use the car as a weapon on christmas markets
On the contrary, autonomous vehicles (especially if something like automated package delivery becomes an offered service) allow car bombs without the suicide bomber inside. Instead you just place an explosive in the device and send it to the intended destination - as an organization the primary concern then becomes whether the ride is traceable back to you which might be do-able given how "frictionless" rideapps are as long as you get a burner phone, prepaid credit card and are very careful about making sure the device is clean and can't be tied to your person.
Right now the tech is limited to places where carbombings aren't commonplace, but I think driverless cars open a huge potential issue in other markets. I absolutely agree they're safer than drivers are, but let's not be hyperbolic.
> autonomous vehicles (especially if something like automated package delivery becomes an offered service) allow car bombs without the suicide bomber inside
Every city has courier services, including Uber.
I don't think you can just look at the statistics though. What I mean is that, the main issue is that the category of mistakes they make are going to be inherently different from the category of mistakes that humans make. IE, humans are more likely to make emotional mistakes, but self driving is probably more likely to make perceptual mistakes or fail to respond to novel situations. I think the "fail to respond to novel situations" is probably the biggest blocker for people to trust them enough for widespread deployment. I'd love to see one of these things drive on a poorly mapped dirt road up in the mountains, for instance (I'm not saying it can't -- I just mean I would personally love to see how it can adapt to that kind of situation)
> I don't think you can just look at the statistics though. What I mean is that, the main issue is that the category of mistakes they make are going to be inherently different from the category of mistakes that humans make
Would you rather have a X% chance of a loved one dying from a human-category mistake? Or a 0.5X% chance of a loved one dying from a ML-category mistake? I would take the latter in a heartbeat. Why waste time optimizing for "categories" when you can optimize for your loved ones not dying
That is not the way human beings typically evaluate risks. 'Acceptable risks' are more about psychology and not a statistical calculation. A self driving car that is twice as safe as the human average would be considered an unacceptable risk.
Add something that is new/unknown + lack of control, and you will have to clear a very high threshold to be considered 'acceptably safe'. My guess is a minimum of a 0.1x.
While I agree with you, not everyone does.
You and I are thinking with a purely logical thought process, but most people don’t reason that way.
A lot of people want to be able to hold someone accountable, and that is more important than the pure probabilities.
They want to be able to blame a person for making a mistake.
> humans are more likely to make emotional mistakes, but self driving is probably more likely to make perceptual mistakes or fail to respond to novel situations
Humans fail chaotically. AVs, at least Waymos, fail safely. In urban environments, there is a case to be made within a few years for banning (or making it massively cumbersome) to manually pilot a private vehicle.
That's an odd claim. Human failures are uncorrelated, but the Waymos all running the same software must be expected to fail in the same way, in the situations in which they fail. Which is why they often make the news; one Waymo stuck soon sees many of its brethren join it.
Don't humans fail in the same way too, broadly speaking? Wouldn't it depend on how you categorize failure?
If a Waymo failed because it missed a stop sign, and then another Waymo failed because it was confused about some strange markings on the road, is that not failing "the same way" in the same sense that a person failing to yield or driving drunk?
Statistics is still the most reliable indicator, because you can’t really tell if a situation is “novel”. How novel is novel? I doubt even Waymo knows for sure.
People will trust them by familiarity, not by any objective measures. Their trust grow as they see them more often, without bad accidents.
My understanding is that they only operate in areas they’ve had approved and pre-mapped out for them though. Practically I don’t think the scenario you’ve described would ever happen, but I too would like to see it (I suspect they wouldn’t handle it well).
I live East of DTLA and twice I’ve seen people driving them around, presumably to map those areas out. It’s pretty annoying that I can’t use them for my late night commutes yet, but I’m eagerly looking forward to when I can!
IDK, I've been in the car with someone when they just up and up drove off the road because they failed to notice that there was a corner - and that's a sober human. Another time, someone else, they rounded a corner, saw a tractor in the other lane, and were so surprised that they hit a tree. My mother has totalled six cars - she insists it was someone else's fault every time, but the statistics are somewhat damning. My wife's driving has given me haemmorhoids, I clench so fucking hard. Blind corners at twice the speed limit, looks over her shoulder to talk to passengers, and I constantly find myself buying wing mirrors, fenders, headlights, body panels, hubcaps, and anything else she can break off on traffic, rocks, or street furniture. Well, I did until someone finally took her license away from her for a year.
So... not so sure. Every AV I've ever been in has driven incredibly cautiously compared to a human driver.
> a self-driving car will never … use the car as a weapon
You are lacking in creativity and vision. With networked autonomous vehicles, all cars can be used as a weapon simultaneously.
> meaning that we don't tolerate machine incidents the same way we tolerate human ones.
Any company that builds self driving cars or operates them should show how reliable they think they are. Not by showing pr material or stats, that's all 'free'.
Make yourself, as the company, legally liable. If you think it's good, put your name on it and own the risk.
When that happens, I'll trust whoever says it's safe.
When a human driver messes up, there's a person to hold responsible. When a robot car messes up, who is held responsible? If the human messes up to the point they are no longer allowed to drive, that person is no longer on the streets to cause more mistakes. When a robot car messes up, are all of the other cars pulled off the streets as well until the glitch is fixed? Every robot car is essentially the same driver.
While we hold humans financially responsible for all accidents, we don’t hold them personally responsible in every case. Most accidents end with no arrests or even tickets. If your hand slips on the steering wheel, you don’t go to jail.
Robot cars would be the same. Insurance is financially responsible, and if the accident is deemed caused by a flaw, it would be the company that made them that is held responsible for fixing it.
Why would it work any differently than any other category of product or service? If you’re rich you can sue them, if not you’re fucked.
Unless there’s a very bad and widespread issue. Then you join a class action and get $24 on a visa giftcard you can’t figure out how to redeem and then forget about.
Mistakes aren’t glitches when the environment is so dynamic. I imagine they are making errors constantly, and constantly learning from them.
When a robot car makes a mistake, all robot cars learn from the experience. When a human messes up, only that human learns from it.
I use Old Glory Robot Insurance.
https://www.youtube.com/watch?v=g4Gh_IcK8UM
> When a robot car messes up, who is held responsible?
The company owning and operating the robot?
Ah yes, can’t wait to sue Google on equal monetary and political footing. It will be very easy, and very fair. /s
We don’t tolerate machine incidents because who is liable if an accident occurs? A company with a $4.6 trillion market cap? They’re not a legal opponent any of us really want to take on.
Lawyers are salivating at the possibility of suing Waymo if they ever make a mistake. Much better to be hit by a Waymo (or a Tesla Robotaxi) than the typical uninsured, "judgement proof", human driver.
Exactly, imagine the huge sums of liability insurance Google is carrying to make payouts when the worst-case scenario happens. That financial burden is a prime driver of their operational safety record.
Any stats to back up that the _typical_ driver is uninsured?
An exaggeration on my part. About 16% of all drivers are uninsured (https://insurance-research.org/node/130), but "accident probability of uninsured drivers is 34 percentage points higher" (https://www.tandfonline.com/doi/full/10.1080/19439962.2018.1...).
If a human makes a mistake, we give them the liability to own up to their mistakes, and carry the costs.
Most self-driving enterprises are trying very hard to ensure their non-driver continues to carry this liability instead of their company.
This helps a them v. us attitude. The company needs to be trusted widescale by the public, and do seem to undercut their own PR efforts on that, repeatedly.
> On the other hand we don't give technology the same pass we give to humans, meaning that we don't tolerate machine incidents the same way we tolerate human ones.
It comes down to accountability, IMO. We tend to be forgiven to humans because we are humans too and we know we, humans, make mistakes, and these can have serious consequences.
When a human makes a mistake, they are immediately accountable, and therefore judged and forgiven (or not). When technology makes a mistake, who's to blame? The CEO? The software developer who wrote the code? The tester? The architect?
Yes, we want technology to be an order of magnitude safer than humans for this reason, because we know that, given the current way our judicial system works, when technology makes a mistake, no one is held accountable.
> On the other hand we don't give technology the same pass we give to humans, meaning that we don't tolerate machine incidents the same way we tolerate human ones.
I completely disagree with this. I don't "give a pass" to humans killing other humans with a car.
Every year that skeptics delay adoption of autonomous driving is another year where tens of thousands of deaths could be prevented.
> we don't give technology the same pass we give to humans
This was the hypothesis. But based on e.g. FSD, it doesn’t seem borne out in reality.
We have a much higher level of tolerance for things that benefit us while posing a risk to others ("should my own car have FSD") than we do for things that benefit others while posing a risk to us ("should Waymos be allowed in my neighborhood").
> a magnitude of order
unless I'm missing something, you meant "order of magnitude"