Hacker News Reader

Comment by roenxi | Hacker News Reader

I would like to nit-pick and say you're estimating some novel metric. The value add would be somewhat higher than that because the alternative is a human driver, so the value-add of the software is approximately ($/h value of driver x hours) that it frees up to do something else. Plus all the evidence I see suggests that whatever the current state, sooner rather than later computers are going to be superhuman at driving (like they are at most other discrete tasks) and save a lot of lives. Plus reduced wear & tear on vehicles.

$365B in value is probably a serious undercall. The only reason to complain here is if only $100B has been put in to the venture so far.

View on HN

I think computers will become superhuman at driving under the right conditions.

However, there will inevitably be conditions that require the use of general intelligence (rather than driving heuristics), and in those situations all you can do is pray the computer acts rationally despite not having GI.

I think self driving cars have already passed the test of "number of crashes" or "number of fatalities" per mile driven. But I don't think that's enough to sway the public, if every once in a billion miles a self driving car slowly drives off a cliff for no apparent reason.

Nothing says self driving cars can’t phone a call center for the 0.001% edge cases. Just add a cellphone connection and a Starlink receiver in the roof for a backup. At which point we would need to add some cell reception to some tunnels and cars can have literally global coverage.

High drips the problem drops from near AGI to not outrunning the cars ability to stop without hitting anything.

>I think self driving cars have already passed the test of "number of crashes" or "number of fatalities" per mile driven.

This is not definitively known. The distribution of conditions under which self driving cars operate is very different from the distribution of human driving. Self driving car miles are disproportionately on the highway, with little traffic, in perfect weather (i.e. by far the safest driving conditions). In addition, we don’t know how many disengagements (or remote interventions) would have resulted in an accident.

True, true. However, if I had to bet... I'd bet in a fully self driving world (using today's tech) there'd be overall far fewer fatal accidents. However there'd also be a lot of bullshit we are simply unwilling to deal with (e.g. cars going 10mph in the rain; slow motion fender benders; car unwilling to turn left; gridlock at intersections; lots of random oddities). I guess my point is that safety metrics are certainly important stats to gauge self driving progress, but certainly not the only metrics that matter, and perhaps not even the most important.

I think that’s underestimating just how bad people are at driving. There is little meaningful difference for other drivers from a bunch of self driving cars suddenly blocking traffic and an actual traffic accident.

People aren’t going to trust cars to safely do level 5 if they can’t do level 4 99.999% of the time. So sure there will be occasional stories of i95 blocked for 3 hours due to software bug, but how is that different from a major accident that occurs regularly?

This is the argument self driving car companies are currently trying to make. My contention is this reason doesn't seem compelling to the general public. For whatever reason the public seems more tolerant of human error than machine error. At this point anyway. I foresee a long period of trust-building before widespread adoption.

An interesting experiment would be for Uber to send two cars for pickup, one human and one self driving. And let people choose.

That’s fair. I suspect we are at least 30 years from a drunk to be legally able to order their self driving car home without issue simply due to cultural inertia.

Assuming we get to that point it will probably be another 20 years after that before non self driving is seen in the same way as driving a motorcycle is today. Aka something that’s not suicidal, but defiantly excessively dangerous.

I'd agree with that estimate. But sheesh, 50 years from now seems like forever.

Two final thoughts...

(1) Maybe within the next 50 years devs can instill into AI, some meaning of death. As it is, I find some comfort knowing my driver realizes the difference between a field and a 40-foot cliff along the coast of Big Sur, and our shared theory of mind regarding the consequence of swerving to avoid something in those situations.

(2) Regarding humans being more tolerant of human error. I think this might be because when a human gets in an accident, there is always the ability to reason that person is different than us: old, tired, drunk, distracted, etc. And both the situation and the cognition are unique to one person. Naturally, we would have done something different to avoid the accident, we reason. If an AI gets into an accident, and we know that same exact AI is driving 10 million cars, including our own, that freaks us out a little.

I think (2) is more cultural than that. People reacted poorly to automatic elevators when they where first introduced and now people just don’t think about them as much more complicated than a light switch. The elevator broke and my light broke have the same feel of an inanimate object not working even though elevators are a lot more complicated.

I'm not sure if "the breaks on the train failed" and "the train's ai suddenly decided to start running in reverse" would get the same reception even if the result was the same.

Yea that feels different. I kind of want to say odd accidents definitely get attention or something, but I don’t have anything to back it up.

Yet aircraft autopilot fails, pilot error, and mechanical error seem to all get the same attention. That might be because the pilots are also at fault for autopilot issues, or it could be autopilot used to be really really dumb so there was a lot of stories that autopilot flying into a mountain etc to prep people with how dumb it was or it could be something else.

If that makes sense?

I actually do feel more comfortable with the idea that autopilot is landing the plane rather than a human pilot. Especially at night. And I don't even know if it actually works like that. Interesting. Good chat.

>>Plus all the evidence I see suggests that whatever the current state, sooner rather than later computers are going to be superhuman at driving (like they are at most other discrete tasks) and save a lot of lives.

What is that evidence, exactly? I agree that we might eventually get there, but the scale seems to be 50-100 years at this point. We are as arrogant as the researchers in the 60s who famously announced that absolutely perfect image recognition is only 1-2 years away - except the problem is several orders of magnitude harder.

Typically from what I've seen computers go a long period being hopeless at a task, then slightly subhuman-par-superhuman very quickly. It seems to me that as far as algorithm-and-Hz cares human intelligence lives in a very narrow window and once computers get close to it they tend to jump over it.

I'd judge self driving to be slightly subhuman right now - there are definitely worse drivers on the road (typically impaired - drunk, near-blind or high). I'd expect superhuman performance this decade just based on that and the rate of improvement in anything AI related right now.

Reply on news.ycombinator.com