[bunderbunder]

I could see there being a nice "hybrid" avenue, too: Have a human driver in the truck for managing all street level driving, overseeing loading/unloading, refueling, etc. And then let them take a break, or work on their side hustle or master's degree or whatever, while the truck is on the highway. And either pull off the highway or have the human take over if conditions aren't favorable.

Self-driving taxis, OTOH, feel like they've got a much longer way to go before they can generate any real profit.

[privacypoller]

This is generally seen as the path to full automation - you can couple autonomous "easy" highway driving with a remote driver taking over for the final mile. How long until we have aircraft doing the same thing ? ;)

[0xfeba]

Well once airborne, aircraft already fly themselves and do everything except the last 100 feet of landing already.

[brudgers]

Aircraft have six degrees of freedom. But for the ground, birds, weather, and other aircraft there is nothing a plane can run into. At 20,000 feet that mostly reduces to other aircraft. And still humans on the ground orchestrate among the flights and provide specific direct oversight of each and every flight in real time. Watching for weather. Watching for birds.

Don't get me wrong, I find aircraft automation impressive. But there is a massive human workforce that makes it possible for the cabin crew to run planes on autopilot. There's a mountain of rigid regulations, licensing and certifications that control every part of that workforce. Every part of each aircraft. Every piece of communication.

That's not how the roads work.

[Tomte]

And for avoidance of aircraft: except for some last-second emergency reaction, this is the opposite of automated. We have specially trained, highly competent people on very short shifts under pretty much ideal working conditions ensuring that.

We just don‘t place them inside the aircraft.

[slavak]

We place them _both_ inside the aircraft and on the ground. While ATC does provide separation, conflicts still can and do happen. Aircraft have onboard systems to warn of conflicts and even to suggest corrective action, but the pilots must be the ones to make the correction.

[mrguyorama]

Indeed, I believe TCAS is one of the only places where planes and pilots override ATC, in that if ATC tells you to go down, and TCAS tells you to go up, you go up

[foota]

In fairness, cars can stop or divert in a couple of seconds. Planes, not so much.

[geezerjay]

> Aircraft have six degrees of freedom. But for the ground, birds, weather, and other aircraft there is nothing a plane can run into.

I would also add that aircraft are monitored by ground control stations, while cars are controlled by the driver alone.

[greenyoda]

> Well once airborne, aircraft already fly themselves and do everything except the last 100 feet of landing already.

That's only if nothing goes wrong. On the other hand, if the airplane I'm in loses both its engines and has to land on the Hudson River[1], I'd much prefer to have an experienced pilot and copilot in the cockpit.

[1] https://en.wikipedia.org/wiki/US_Airways_Flight_1549

[Piskvorrr]

Black swan events are not very useful points of comparison - this event is not dubbed Miracle on the Hudson for nothing.

[astronautjones]

With constant, vigilant monitoring

[kosievdmerwe]

For some definition of constant, planes aren't seconds away from disaster like cars/trucks are pretty much constantly.

[Piskvorrr]

Since the context here is "autoland", yes, that means "do not stop monitoring the bird, be always ready to take control." Landings are one of the trickier parts: you are moving at hundreds of mph, on an almost-collision course with the runway, by definition: the plane is supposed to stop flying very few feet above the concrete, AND not drop too hard.

[the_duke]

Actually, most modern passenger aircraft can land autonomously on properly equipped airports with CAT 3 autolandings.

(Note that it's not fully autonomous, the pilots still need to do quite a few things like lower flaps, extend the landing gear, ....

[IshKebab]

Yeah he knows.

[noodle]

AFAIK this is the accepted path forward (minus the side hustle aspect). Automate that which is easier to automate: the long distance highway driving. For the first/last mile, let a human do the work.

[colechristensen]

I think what is going to happen is people are going to realize robot cars are death traps. Not because their rate of accidents will be any higher, but that when they do have accidents they will seem bizarre and inhuman mistakes that even the most incompetent human driver would never make.

Driving on the open road requires real intelligence. Not the pretend intelligence that modern AI gives, but real understanding of situations and terrain. Before that happens (which is basically skynet, and a very very long time away) all you have is a bag of tricks cobbled together. Those tricks will miss things and get confused and make mistakes. Maybe not very often but definitely in strange ways that are frightening.

The unknown is scary. Drunk drivers, tired drivers, old drivers, et al. are plenty dangerous, but they still behave in ways that can be understood. AI mistakes will be / are / have been strange unsettling things that can't be reasoned about if you're a person in the area of the misbehaving vehicle.

[philwelch]

> Driving on the open road requires real intelligence. Not the pretend intelligence that modern AI gives, but real understanding of situations and terrain.

People used to say this about every single thing that computers can do better than people.

In my college town, some pedestrians got ran over by a driver who later pled insanity due to "caffeine-induced psychosis". I think you're seriously overrating the predictability of human failure modes.

[emodendroket]

> People used to say this about every single thing that computers can do better than people.

But computers can't do it better than people! That's what drives me nuts about this debate -- it's just accepted as a premise that either the self-driving cars are much safer than human drivers, or the path to getting them there is very close and no serious obstacles remain. Neither is true and it's not clear they will be. https://blog.piekniewski.info/2017/05/11/a-car-safety-myths-...

[baddox]

I think what is clear is that virtually no one is okay with large-scale public deployment of self-driving cars that aren’t clearly statistically safer than human drivers, so when we’re talking about public deployment of self-driving cars, it’s implied that we’re talking about when (if) they reach that level of safety.

[emodendroket]

Frankly, that's not clear to me. A lot of people seem quite eager to put the first vaguely plausible thing all over the road because of an exaggerated idea of the incompetence of human drivers (which is understandable, but harmful in this context)

[philwelch]

The question is, which human drivers?

Let's say we have a self-driving car that is as safe as the 20th percentile human driver. Do we allow that self-driving car on the roads? Do we selectively revoke licenses from 1 out of every 5 drivers and replace them with a car that's at least as safe as they are if not probably safer? Do we replace breathalyzer interlocks for drivers with DUI convictions with an AI driver and just revoke their licenses permanently?

There isn't a trivial solution to this problem. At some point, some AI driver is going to cause an accident that would not have been caused by a 95th percentile human driver. At the same time, human drivers do shit like this all the time: https://www.youtube.com/watch?v=oidHSzukSss

[darpa_escapee]

To me, this seems like an a priori assumption that accepted as the truth. Then, when it's questioned, the person questioning is made out to be some kind of luddite.

Which is an odd phenomenon to me. I can't even get Google Assistant to understand me 3/4 of the time, yet I'm supposed to take it on faith that autonomous cars are inhumanly safe?

[lucideer]

> People used to say this about every single thing that computers can do better than people.

Did they though? Think about things that computers can do better than people: they're mainly things that we completely predicted computers would be better at (arithmetic, precision manufacturing, drafting, telecommunications routing). Beyond that, you're left with things computers are only better at dependent on priority, the canonical example being service jobs where economics trump's QoS; computers are much worse than a cashier, but comparatively cheaper by a margin that makes the quality compromise worth it.

The only possible exception I could think of that's come up in recent discourse is diagnosing patients, but even that, while encroaching on a role that has traditionally been revered as a career, is still something that seems at least on the surface to be quite predictable given the nature of what's required to make diagnoses (simultaneous access to a trove of data and knowledge).

Beyond the above, I think it's pretty reasonable that there's a broad range of things computers will not be better than humans at for a very very long time, if ever.

[baddox]

But this is a cliche in the AI field: that AI is defined as the things that computers can’t do as well as humans right now. As soon as computers match human ability, well, clearly that doesn’t count as “intelligence.”

And I don’t think the problems computers have proven themselves useful in solving have been what most people expected. Chess, Go, facial recognition, Jeopardy, image classification (hot dog or not), captchas (clearly, since they’re designed specifically to resist computer solutions), etc. seem to me to be things that, before computers proved to be decent at, would have been widely considered to require intelligence on the level of humans.

[foldr]

People only thought that chess required intelligence because they had no idea how much raw computational power computers would obtain. To anyone who understands the rules of chess, it's obvious that a machine which can perform a near-exhaustive search of the state space for a few moves ahead is going to be able to play chess better than a typical human.

[wyattpeak]

That just isn't true. It was widely believed through the seventies and eighties that chess inherently required creativity, and a machine could never beat a grandmaster.

Rather, I suspect, tasks which computers start outperforming humans in we reanalyse as "completely procedural". Nobody called chess procedural in the middle of last century.

[gaius]

Well, a computer playing chess is simply enumerating all possible boards. Which is the same way that a GAN produced “art”. Humans do it creatively because that’s how you do it if you lack exhaustive computing power and memory. In neither case is a computer mimicking that process of a human, they simply arrive at the same outcome by a brute force means.

[rayiner]

> People used to say this about every single thing that computers can do better than people

Even if that’s true (and I doubt it is), there is ample precedent (AI winter) for the industry dramatically overestimating what computers can do.

I bet if you time traveled and showed Siri/Cortana to an AI researcher from 1960 they’d be incredibly disappointed.

[pcwalton]

> I bet if you time traveled and showed Siri/Cortana to an AI researcher from 1960 they’d be incredibly disappointed.

It's a common misconception that the 1960s and 1970s were a time of unbridled enthusiasm in AI. In fact, there was a ton of pessimism back then too: for example, ALPAC [1] was so pessimistic about the future of natural language processing that it got the US government to pull most of its funding.

I think if you were to show Siri, Alexa, etc. to some of those folks they'd be pleased that we've gotten as far as we have, while acknowledging the obvious fact that there's plenty more to do.

[1]: https://en.wikipedia.org/wiki/ALPAC

[ghaff]

Not sure why people are disagreeing. That seems a blindingly obvious comment. There's so much today that would seem almost like magic to pretty much everyone living in 1960. But coice assistants? (And even just voice recognition.) almost certainly seemed like relatively "easy" problems. Perhaps less so to AI researchers than the general public but still.

[rayiner]

I don't know about "magic." A modern smart phone might be "magic" to someone in 1860--it operates based on technologies that didn't exist back then. But by 1960, the building blocks of modern computing were already in place: digital von Neumann computers built out of transistors, radio communications, signal processing, etc. AT&T used frequency-division multiplexing of multiple voice channels in phone transmissions in 1918. The mathematical framework for modern technologies like LTE was in place by the 1950s and 1960s. Would it really have surprised anyone that transistors would continue to get smaller and faster, allowing higher complexity, higher-throughput signal processing, which would allow Facebook?

[blacksmythe]

  >> if you time traveled and showed Siri/Cortana to an AI researcher from 1960 they’d be incredibly disappointed.

If you time traveled and showed 2013 me what self driving cars would be doing in 2018, 2013 me would have been astounded.

[YeGoblynQueenne]

>> People used to say this about every single thing that computers can do better than people.

Who were those people, who said those things (i.e. where they AI researchers, or computer scientists?). And what exactly did they say?

There have always been strong criticisms of AI (e.g. [1]) and opinions dismissing computers voiced by people who did not have an adequate understanding of computers.

The interesting thing is to see what the people in the know actually thought over the years and what they think right now.

Edit: to clarify, what AI researchers usually do is overhype the capabilities of their systems and claim they can achieve things that they never manage to show they can- completely the opposite than saying that "computers can't do that".

__________________

[1] "What computers can't do" by Hubert Dreyfus

https://en.wikipedia.org/wiki/Hubert_Dreyfus%27s_views_on_ar...

[a3n]

> People used to say this about every single thing that computers can do better than people.

And they were right, until they were eventually wrong. There will be this phase for automated vehicles too.

[Piskvorrr]

True. However, "eventually" is a long time - the assumption "level 5 hardware exists today" is IMNSHO premature by many decades.

[mehrdadn]

> when they do have accidents they will seem bizarre and inhuman mistakes that even the most incompetent human driver would never make.

I agree -- but it's even worse. Even if robots do make human-like mistakes, that doesn't mean humans will be forgiving of the same mistakes. For one thing, I might forgive a human being unable to react due to a 1/2-second reaction time, but I sure as heck wouldn't be that forgiving for a robot. I would expect and demand an order of magnitude better. For another thing, people have more tolerance for mistakes made by "closer kin", if you will. (e.g. if my own child steals from me, even 10x as much as a random thief does, that doesn't mean the thief can expect more lenience than I had for my kid.) Self-driving cars pretty much have to be strictly _and_ significantly better than more than the majority of humans for people to trust having them around. Merely being better than average, even if it's in all respects, isn't necessarily enough to cut it.

[jlebar]

> I think what is going to happen is people are going to realize robot cars are death traps.

Drop the "robot" -- it's cleaner.

[Piskvorrr]

While that's true, the debate rages around "the devil you know x the one you don't." Death traps on wheels are known (and risks ignored), their autonomous modes...not yet.

[verelo]

The funny thing here is we might already have the best solution to the "easy" automation scenario: Trains. It's a shame they're so expensive and require special tracks. If we could invent a hybrid train / car, that behaved like a train, but didn't require special track...that might actually be a better direction.

[ghaff]

You're pretty much describing a truck or a bus in that breaking up a train is one of the things you need to do to allow it to travel on non-dedicated/special right-of-way.

And, with respect to freight, a lot of the easy automation is handled by trains. A huge amount of truly long distance freight in the US (including but not limited to bulk cargo) goes by train for much of its overland transport.

[Piskvorrr]

The special/segregated track is one of the major reasons they're more secure.

[TylerE]

Aussies have been doing this for decades.

https://qph.fs.quoracdn.net/main-qimg-2bc514ba621d41419fabd6...

[conquistadog]

People seem to have much greater fear of spectacular and bizarre deaths than of considerably more likely but relatively mundane deaths.

[tigershark]

Why are you so sure? Waymo demonstrated that they can handle correctly a duck-chasing wheelchair rider that used a broom. I’m not sure that all the human drivers would have handled it so well.

[Piskvorrr]

And Uber demonstrated that incorrectly classifying a detected obstacle is solvable by running it over. I'm pretty sure that no sane human driver would have handled "could be a bike or a human or an animal, but it sure is something in my path" by ignoring it for six seconds before the collision and even afterward.

"But we should let it out on the road, it drives on par with an insane, legally blind and completely drunk driver" is not a very convincing proposition.

[kozikow]

> Automate that which is easier to automate: the long distance highway driving

Long distance highway driving except that one highway exit on 101 that kills you. I'm just saying that when human lives are at stake, bottoms up approach may not be as feasible as with web software.

[noodle]

> except that one highway exit on 101 that kills you

The conversations on the topic that I've heard involve building special, dedicated exits, sort of like truck weigh stations. And having them live not particularly close to urban centers.

So it a bit more of a holistic approach than bottom up, and something that will take a while to implement since you're not able to roll it out everywhere at once.

Not to say that there still won't be issues. I'm sure there will be. And I'm sure there's a long way to go still. But its also not a black and white issue.

[slfnflctd]

At this point I'm starting to think engineers for self-driving vehicles need to find a way to explicitly map out every possible scenario on a given stretch of road... and from that - in real time - derive a custom template for each individual vehicle, tailored to the range of possible interactions between the vehicle in its current state (i.e. tire pressure is significant) and the current scenario playing out on that road.

Much of the process of getting close to this point can be done with the latest A.I. tools, but I have this nagging feeling we're going to need humans to fine tune a whole lot of 'last mile' stuff.

[nradov]

Unless you ban all human drivers from that road, and find a way to keep out pedestrians and animals, there will always be an infinite number of possible scenarios. Mapping everything out in advance can't possibly produce the level of safety that the general public would demand.

[myrryr]

But with driving human lives are at stake now.

It just has to be better than the average driver.

[ultraluminous]

That's one more of those "irrational exuberance" regarding the future of autonomous driving. It would never even come close to being legal if it's only better than the average driver. It will need to be better than the 95th percentile of human drivers at the very least. 48% of the human population are not going to accept a more dangerous, potentially deadlier, car ride because it's "safer on average".

[slavak]

In which case any driver of above-average ability would be insane to put their lives in the hands of one.

How many people do _you_ know that consider themselves below-average drivers?

[michaelt]

I'm just spitballin' here, but perhaps they could make something that was better than the average driver, and wouldn't swerve or accelerate into stationary objects like highway barriers? That sounds like something a lot of people would want.

[Piskvorrr]

Looks easy, right? Too bad it's been looking easy for 50+ years now, with no solution in sight.

[Angostura]

You could have 'Pilots' like you do for the last part of sea voyages

[gaius]

The model of truck to the railhead, rail for long distance, the back onto trucks for the final leg already exists. Trains have large economic advantages over trucks for well known routes. So what are the advantages of self-driving trucks in this scenario?

[sitkack]

> master's degree or whatever

Are you trying to make people make fun of us?

[goatherders]

Dont be so cynical. I took the comment as serious and agree.

[toomuchtodo]

Side hustle or masters degree? Pilots don’t work another job while autopilot is in use, and neither should truck drivers monitoring its autonomy system.

[loeg]

A self-driving vehicle would be a step beyond current autopilot, and asking any human to constantly monitor an autonomous system is going to result in pretty catastrophic failure. An autonomous system needs to be fail-safe to be used.

Plane autopilot only works because air is so empty and flying in the same direction at constant elevation is unlikely to result in any problem. There are repeated examples of both pilots falling asleep and planes over-shooting destination airports, for example.

[toomuchtodo]

> A self-driving vehicle would be a step beyond current autopilot, and asking any human to constantly monitor an autonomous system is going to result in pretty catastrophic failure. An autonomous system needs to be fail-safe to be used.

I think the idea we’ll attain this in the next decade or two is borderline delusional, but to each their own.

Commercial aviation is one of the safest transportation mechanisms in the world; aircraft can go runway to runway in mostly automated fashion (auto throttle/TOGA [take off go around] for takeoff, autopilot for cruise, autoland for landing). We (customers and regulators) still require human attention the entire time.

[travisjungroth]

> aircraft can go taxiway to taxiway fully automated.

Source?

[toomuchtodo]

My comment was unintentionally overgenerous. I’ve corrected my statement after reviewing the aviation stackoverflow site, and flushed it out with more details (specific aviation terms).

[bunderbunder]

If it requires human oversight, what's the point?

At that point you're getting no more bang for your buck, since the operator is going to be subject to the same limits on time behind the wheel as a driver of a non-autonomous truck. And you're not getting any more safety, because, as Uber and Tesla have been illustrating for us so vividly, a self-driving system that needs a human overseer can't drive safely, and a human who isn't physically in control of the car at all times can't oversee safely.

(Edit: This is, naturally, not accounting for the need for a transitional period while getting the technology bootstrapped. But that's time invested in developing the tech, not time where the tech generates any profit.)

[toomuchtodo]

> At that point you're getting no more bang for your buck, since the operator is going to be subject to the same limits on time behind the wheel as a driver of a non-autonomous truck.

Human lives and property loss are expensive. The average cost of a fatal crash is well over $3 million. The average cost of a large truck crash that does not involve a death is approximately $62,000. Settlement payouts due to big rig accidents are roughly $20 billion per year.

You don’t need to replace the driver to see significant upside.