We like the idea of technological solutions that don't require us to change our lifestyle or make hard choices. Driverless cars are appealing because they would allow us to largely maintain the status quo (as opposed to, say, reorienting toward public transportation). It's like an obese person who says, "I don't want to eat healthy and exercise. Why can't scientists just invent low-calorie versions of the food I like?"

Really? I don't like to fetch water from a lake every day to my house. Instead of using technology to bring water to my house, what lifestyle changes or hard choices should I rather make?

The great invention for dealing with this is a public utility that provides water to everyone. A mass transit system for water.

But why not a fully-autonomous robot that employs machine learning to detect nearby lakes and streams that can hand-deliver the water?

Tuche!

The bigger problem, IMHO, is the attitude that it's some kind of moral quandry if technology allows us to solve a problem without effort. For example, I've heard people really say that an obesity-preventing drug would be a bad thing because it would discourage exercise. This fetishization of hardship does a lot of harm, because in reality, the kind of progress we can make as a civilization essentially amounts to refocusing exertion on increasingly impactful things.

Other examples of this phenomenon:

Direct atmospheric carbon removal is bad. It encourages emissions!

Contraceptives are bad. They encourage sexual license! (I had to try to avoid the obvious joke here.)

Highways are bad! They encourage driving.

Word processors are bad. They discourage good penmanship!

Fertilizers and farm equipment are bad! They discourage small family farms!

GPS navigation is bad! It discourages map reading and mental direction finding!

Some of these examples look ridiculous to us now, but people really did think these things at some point. I see some opposition to autonomous vehicles falling into the same category.

Okay, what is the allegedly obvious joke about contraceptives?

They too encourage emissions.

We like the idea of technology improving our quality of life because as a civilization it's worked for us spectacularly so far. Driverless cars are appealing because they build on top of a transportation solution that already works, reducing the problems and externalities of it, without requiring us to completely overhaul the way we build cities and live which could take decades. When half of the world lives in remote or insecure areas, driving vs public transport is not a lazy lifestlye choice like eating poorly. Autonomous cars will also bring about a lot of new use cases which could improve public transport design, reach and usage, like driving you to a metro station which has no parking in the morning and picking you up in the afternoon.

And as a society we reject a lot of possible technology out of hand.

For example, we could have a device on cars that transmit basic information. Such as licence ID number, speed, location, pedal position etc. It would be ridiculously useful data for a whole host of different things. You could use it to bill parking, catch speeders, sequence traffic lights, and warn of hazards ahead. And maybe even help self driving cars make sense of the world.

From a technological point of view this is a no-brainer. But this is incompatible with the way that we think about cars and roads. Every car has to pretend to be isolated from every other car.

> From a technological point of view this is a no-brainer.

Yeah, it becomes a brainer, though, when you ask “how do you deal with malicious transmitters”, since the answer to that tends to involve developing the systems to monitor the same variables externally without trusting the transmitted values, and once you do that, what is he transmitted data for?

Using transmitted data for less safety-critical things like route intent for efficient holisitc route optimization, with a penalty system for misrepresentation might be useful, though.

I think that safety critical systems need to deal with bad data anyway. Adding an extra level of redundancy makes sense.

Maybe you could have a lower power transponder style technology in addition. This would just broadcast an ID after a fixed amount of time in response to a request. This could be compared against radar,visual, and broadcast information to actively verify data. And critically it could be independent of the rest of the cars systems. Obviously that is not a solution, but it does add another layer of redundancy

Sounds preferable in both cases to me What is the non tech solution for automobiles, better designed cities?

Sounds to me like driverless cars are easier than redesigning existing cities.

Redesigning cities to be less reliant on cars is pretty much a solved problem. It just needs a few decades of focused policies. Check out for example how the Netherlands became bike friendly by relentless improvement since the seventies.

And being completely flat and never getting very hot/cold. Bikes aren't really workable in most cities as a primary form of transit.

Ebikes make most cities flat, and the Netherlands only relies on bikes for last mile transport. Medium and long-haul routes in the Netherlands have great transit options.

But human drivers kill a lot of people. It would be a shame if that's the best we can do. And some car tech improvements have saved lives, so it's not necessarily chauvinism it's just extrapolation.

As long as the U.S. personal-injury legal system remains in place (with strong support from media coverage of fatal crashes), it won't be enough for self-driving systems to have lower death rates/mile in aggregate. They will need to be better or equal to human performance in every subcategory of driving. That's really hard.

To wit: It's not enough to say "Our self-driving cars avoid 14,000 drunk-driving, texting and asleep-at-the-wheel fatalities," if it's also true that: "Our self-driving cars hit and kill 30 errant pedestrians a year that a human driver would have noticed."

I know that 14,000 > 30. But the specter of roadway martyrs being murdered by killing machines needs only a few examples to sustain itself.

Tough problem to solve.

They don’t need to match in every category, they just need to match in one category and then detect when they’re in that situation.

It will be a gradual transition as they learn how to handle more and more situations.

Self-driving cars will also kill a lot of people. We just don’t know how many yet.

It will very likely kill far fewer per mile driven than humans do. Unfortunately you're right -- it probably still will kill some. Even nonfatal accidents will be scrutinized heavily and leave a lasting impression of fear/loss of control.

I'd wager a relatively small percentage of drivers account for a relatively large percentage of deaths. Reckless driving, not driving as conditions permit, intoxicated drivers probably account for the vast majority of road fatalities, either directly or indirectly.

I don't believe a computer will ever be as good a driver as I am in my lifetime in terms of my personal safety, except for when I'm really old and possibly can't see well.

That's totally unproven, the technology doesn't even exist yet and we're still assuming we know what it will look like. In their current state self driving cars would kill far more people than humans if all cars used self driving 100% of the time, and no one knows if we will ever move past that.

Despite the imperfect technology, I'd bet that if we could somehow make all cars autonomous today (with current tech) and network them all together, it would drastically reduce both accidents and fatalities.

The vast majority of accidents by human drivers are caused by basic neglect, not tricky edge case situations. I think it's hard to really comprehend just how bad humans are at driving. It wouldn't take anything close to a perfect autonomous system to do better--just having "drivers" that actually pay attention to their surroundings all the time would be a massive improvement, even if there were occasional bugs and glitches.

> In their current state self driving cars would kill far more people than humans if all cars used self driving 100% of the time, and no one knows if we will ever move past that.

Really? No one knows if we will ever improve on self-driving cars than the current technology today? I'm going to stick my neck out and say "Yes, I know that will move past that."

The current state of self driving cars is transient. That much is proven. They aren't getting worse.

The current disillusionment is a natural consequence of the preceding over-hype and over-exuberance.

We've been there before.

After the AI winter, many people argued AI will never come, and many achievements we already have (such as speech recognition) are impossible.

I wouldn’t say speech recognition is solved. It’s still quite error prone.

Self-driving cars need to be even better or people will die.

Perhaps self-driving cars can be deployed in a limited way, as taxis, for example.

>I wouldn’t say speech recognition is solved. It’s still quite error prone.

Sure it is, but it's doing a hell of a lot better than many critics thought ever possible.

>Self-driving cars need to be even better or people will die.

People will always die. Requiring with certainty that self-driving cars will never cause accidents is a purely emotional instinct with no regard for facts or well-being of people. The only thing that self driving cars need to demonstrate is statistically lower deaths or injuries caused per mile (or hour) of driving than the average human would cause. This would mean that adoption of technology would reduce total deaths if substituted for human drivers, which is what actually matters.

>Self-driving cars need to be even better or people will die.

He is right - the problem with your response to this statement is that it doesn't identify the _actual_ reason behind why they need to be better, and it's a problem identified by the article in question: When a human driver swerves randomly and kills someone, we know EXACTLY who is to blame, and something can be done. If a driverless car, or even a car with "driverless technology" that also has a driver in it (like the Tesla driver mentioned in the article) hits someone and god forbid kills them, we still have no idea who is to blame for that problem, because it practically becomes a marketing argument in some cases.

Before we can be okay with driverless cars, ESPECIALLY if we don't plan on them being 100% accident free, we need to solve a lot of ethical and legal battles that don't seem to be even on the minds of the people that tout this as the future, or a necessity.

You argument is basically that it is more important to figure out who to blame for deaths, than it is to get rid of those deaths in the first place.

That strikes me as weird.

Also, the people who will be paying out damages will be exactly the same as the people who are paying out for them today. The insurance companies.

The simple solution to these "ethical" issues is for everyone to have insurance, and for the insurance to pay out.

And I am sure that the insurance companies will be very happy with this arrangement, as reduced accident rates mean that they make more money.

Nobody will accept liability for actions outside of their control, and car manufacturers can not possibly afford to insure all cars they sell in perpetuity.

Insurance only works if the insuree bears the risk (and it has to be an actual risk with nonzero probability times cost, i'll leave why as an exercise to the reader) and in the case of car accidents, that risk is liability, which people have and computers don't. Can TSLA or Uber be charged and convicted of vehicular manslaughter?

Lower rate of accidents will just lower the market rate of insurance contracts.

> Nobody will accept liability for actions outside of their control,

Sure they will. People pay for insurance now, and the insurance companies will give you a discounted price for your self driving car insurance.

Why wouldn't I want to accept liability in exchange for a cheaper policy? I would win as well as the insurance company.

The benefit is clear. And this is an exchange that many many people would be willing to make if it meant that their insurance prices were cut in half. It's free money to the consumers.

You are treating this as if this is some new problem. Your car can already have the wheel fall off, due to no fault of your own, and you'll crash, and the insurance company will pay for it.

> Nobody will accept liability for actions outside of their control, and car manufacturers can not possibly afford to insure all cars they sell in perpetuity.

Car manufacturers already, in the US, bear strict liability for damages resulting from manufacturing defects in perpetuity, and bear the cost of that liability either by self-insurance or some form of liability insurance contract.

> Can TSLA or Uber be charged and convicted of vehicular manslaughter?

Irrelevant to the insurance issue, since criminal penalties are uninsurable; more relevant is can they be held civilly liable for wrongful death resulting from a defect in a vehicle they make, to which the answer is clearly “yes”.

If we look at humans as a counter, sure, apply all the stats you'd like, but I don't think we should do that when it comes to self driving cars. if we simplify the driving risks, we have: 1. driver kills him/herself 2. driver kills person outside his/her car

* Humans and computers can make mistakes on both.

But, my wife and I can teach our kid everything we have learned over the years, to prevent 1, and try to reduce 2. I consider ourselves better than average drivers, if I get my kid a self driving car that is not as good as us, I won't care that it drives better than my neighbor or someone in California.

And in a tragic scenario, it's one thing to say, my kid made a fatal mistake vs, some random QA didn't do a good job and now my family is minus 1

Everyone thinks they are good drivers, yet hundreds of thousands of people are dying every year because of these "good drivers." Not sure what makes you think that teaching your kid what you know will make them a better driver than software that will continue to be iterated on and improved based on millions and millions of miles of data (more data than you and your wife could ever pass along to your child). Also not sure why you think your own driving will be better than self-driving cars.

Exactly. People used to think computers will never beat humans in chess, clinging to all sorts of romantic notions such as that inherently human understanding was required for high-level play.

And now my phone can beat the world champion.

If you care about traffic safety and take steps to be a safer driver (minimize distractions, pay attention, don't drive while under the influence of drugs, including alcohol) then you are almost certainly not a significant part of the problem. The problem is other drivers, whom you have no control over, and who can ruin your life at any moment with sheer carelessness and with few consequences.

By that logic, we should be putting all the money we're putting into driverless cars into getting more people to travel by bus, rail and air, all of which are a much much better than marginal improvement in terms of deaths per person-mile.

You are absolutely correct. That would almost certainly reduce deaths, and I would support taxing myself more for more money to go towards safer public transportation to reduce traffic fatalities.

I will, at the same time, support self-driving car research.

I never use the word certainty, you threw that in there. I said they need to be better than current speech recognition.

Speech recognition can be good, but sometimes it seems to go haywire. It did nail this post though.