Driverless cars

Our transport is heavily oil-based. What are the alternatives?

Moderator: Peak Moderation

Little John
Posts: 8549
Joined: Sat Mar 08, 2008 12:07 am
Location: UK

Post by Little John »

This is precisely this kind of issue I have been talking about. And it is trivial in comparison to the kinds of moral and legal quagmires that await us if this technology is deployed en-mass
User avatar
clv101
Site Admin
Posts: 8771
Joined: Thu Nov 24, 2005 11:09 am
Contact:

Post by clv101 »

1 death in 130 million safe Autopilot miles to date. US averages 1 death in 94 million miles - though these numbers are not directly comparable as Autopilot is only engaged in relatively safe conditions. I have no more of a problem with computer mistaking a white truck for a bright sky as I do a human mistaking a white truck for a bright sky.

"Neither Autopilot nor the driver noticed the white side of the tractor trailer against a brightly lit sky, so the brake was not applied."
woodburner
Posts: 4127
Joined: Mon Apr 06, 2009 10:45 pm

Post by woodburner »

PS_RalphW wrote:http://www.bbc.co.uk/news/technology-36680043

Tesla autopilot 0 Truck 1.
From the link
In a statement, Tesla said it appeared the Model S car was unable to recognise "the white side of the tractor trailer against a brightly lit sky" that had driven across the car's path.
Which is suspect since there are also ultrasonic proximity sensors, so what happened to make them not work either?

As for the video, if the car was really under robot control, why did it not do anything to maintain a safe following distance when the breakdown (or whatever) truck cut across its path?
To become an extremist, hang around with people you agree with. Cass Sunstein
User avatar
clv101
Site Admin
Posts: 8771
Joined: Thu Nov 24, 2005 11:09 am
Contact:

Post by clv101 »

woodburner wrote:Which is suspect since there are also ultrasonic proximity sensors, so what happened to make them not work either?

As for the video, if the car was really under robot control, why did it not do anything to maintain a safe following distance when the breakdown (or whatever) truck cut across its path?
I'm sure the full/real reason for the failure is far more complex than the press release suggests. But that's analogous to human accidents. When someone falls asleep at the wheel and crashes, did he crash because he fell asleep at the wheel? Or did he crash because he fell asleep because he only got 2 hrs sleep the night before because his baby was sick because she picked up a virus at nursery because there was another sick kid who had't stayed at home because his parents etc etc...

I'd imagine a similar chain of events occurring in the Tesla's systems leading to the actual crash.
woodburner
Posts: 4127
Joined: Mon Apr 06, 2009 10:45 pm

Post by woodburner »

Vehicle electronics are bad enough for reliability now, compared to the older mechanical systems, don't let them take over safety critical control until there is an external monitoring system. Neither railways or aircraft (in congested areas) run without an overseer.
To become an extremist, hang around with people you agree with. Cass Sunstein
User avatar
clv101
Site Admin
Posts: 8771
Joined: Thu Nov 24, 2005 11:09 am
Contact:

Post by clv101 »

woodburner wrote:Vehicle electronics are bad enough for reliability now, compared to the older mechanical systems, don't let them take over safety critical control until there is an external monitoring system. Neither railways or aircraft (in congested areas) run without an overseer.
I know Steve doesn't like my view on this, but I'm happy for modern electronic systems to take over safety critical control if they can the demonstrated to perform better than the 'wetware' human equivalent. People screw up all the time, I expect we aren't many years (near-term collapse of civilisation excepted!) from electronic systems, while not being 'perfect', comprehensively outperforming humans.
Little John
Posts: 8549
Joined: Sat Mar 08, 2008 12:07 am
Location: UK

Post by Little John »

My problem is not your extreme utilitarianism. Or, at least, I do have a problem with such a position, but it is at least consistent. My problem is your lack of consistency in that moral and philosophical position (and it is a moral and philosophical position) as evidenced in previous exchanges in this thread. Either you are a hypocrite on this or you simply haven't thought it properly through in terms of its wider moral/philosophical implications. Since you are not stupid, I suspect the former.

What your position is very indicative of, however, is a certain mindset common amongst academic engineers and some scientists in that you presumably view all problems and their potential solutions in morally neutral, mechanistic terms.

Orwell understood that mentality well enough. And so do I.
Last edited by Little John on Fri Jul 01, 2016 11:54 pm, edited 1 time in total.
User avatar
Catweazle
Posts: 2621
Joined: Sun Feb 17, 2008 12:04 pm
Location: Little England, over the hills

Post by Catweazle »

Little John wrote:This is precisely this kind of issue I have been talking about. And it is trivial in comparison to the kinds of moral and legal quagmires that await us if this technology is deployed en-mass
I can only see these systems working if the cars occupant assumes liability and is covered by his insurance. If the guidance computers are genuinely safer than human drivers the insurance companies should be quick to accept them. A bonus from the insurance companies point of view would be a computer record of the events leading to a claim.

My guess is that the computer systems will have similar architecture to the bank transaction processing systems, two identical systems working in parallel with a third comparing the results and flagging an error if the two don't agree.
woodburner
Posts: 4127
Joined: Mon Apr 06, 2009 10:45 pm

Post by woodburner »

Bit of a bugger if the two agree, and are both wrong. It has happened in airliners, so don't think it won't happen in cars.
To become an extremist, hang around with people you agree with. Cass Sunstein
trevorpoole844

Post by trevorpoole844 »

These could be great for people who work so hard during the day and can relax on their way home without driving.
vtsnowedin
Posts: 6592
Joined: Fri Jan 07, 2011 10:14 pm
Location: New England ,Chelsea Vermont

Post by vtsnowedin »

trevorpoole844 wrote:These could be great for people who work so hard during the day and can relax on their way home without driving.
Especially if you have unwound a bit to much during happy hour at the bar conveniently situated between your work place and your home. :roll:
Pepperman
Posts: 772
Joined: Sun Oct 10, 2010 9:00 am

Post by Pepperman »

Some thoughts:

I doubt AVs will have much in the way of trolley problem-based ethical decision making logic designed into them. I suspect they'll be designed to protect the driver and that's pretty much it. When things start to go wrong it's really a case of stopping the vehicle as quickly as possible (AVs will be better at that than humans, that's for sure) rather than come up wtih some elaborate avoidance strategy that involves sacrficially colliding in a way that minimises harm according to some complex algorithm.

But assuming they can actually get them to work in all conditions (and they're still a long way from that in spite of the hugely impressive progress made in the last half decade) then they should be far safer than human drivers so it's not really a major issue I reckon.

Personally I think that AVs could ultimately be beneficial if the result of moving over to them is the end of private car ownership. If that doesn't happen then I suspect it's going to be a disaster, with just the same number of vehicles and a lot of empty running (either heading off home to park up, doubling the mileage, or even worse pottering around while they wait for you to call on them again), probably wiping out all of the benefits that electrification of road transport could bring. The only way I can see for it to work is if people call on pooled AVs when they need them and then the vehicles scurry off to move someone else nearby.

But then we have the problem of what to do about the rush hour when we have as many as 2.5 times as many passenger car trips starting as on average. I reckon something like a third of the fleet would only be used to cover the rush hour - I can't see how that kind of utilisation would be economical. Also the flows are (obviously) heavily biased from places where people live to places where people work in the morning, the opposite in the evening. This means that AVs would be empty for approaching 50% of the rush hour as they reposition. That's not a very efficient use of their time or energy. At least with dumb cars they're only driving as far as the driver needs to drive. Similarly on Saturdays we have nearly a fifth fewer car trips than on weekdays and on Sundays a third fewer.

I like this remake of the classic urban transport comparison:

Image

This is my main problem with AVs in an urban context. The breathless hype ignores the fact that, in Europe at least, public transport does the bulk of the heavy lifting in cities and will continue to do so even if AVs are successful.

Where I do think AVs have interesting potential is in rural areas where they could be hugely beneficial, possibly through regular colectivo (taxi share) styles of service as well as full on demand services.
User avatar
emordnilap
Posts: 14621
Joined: Wed Sep 05, 2007 4:36 pm
Location: Hou�sʇlʎ' ᴉʇ,s ɹ��llʎ uoʇ �oɹʇɥ ʇɥ� �ɟɟoɹʇ' pou,ʇ �ʌ�u qoʇɥ�ɹ˙

Post by emordnilap »

That's a fairly good summation, Pepperman.

I'd come down on the utilitarian side of LJ's hypothetical situation, meaning that if road deaths can be halved using AVs, then there's a strong moral case to introduce them.

LJ, your suggestion of euthanising children with congenital defects, we're not far off doing that anyway, using technology. Not specifically euthanising, more preventing such children being born - preferably not even being conceived - in the first place. Would you welcome such technology (naively assuming it won't be limited to the wealthy)?

Of the remaining deaths, providing AVs do reduce them significantly, the number of cases where LJ's moral choice (post two in this thread) would be invoked is possibly extremely small. In any case, how many drivers would be able to make a true 'moral' decision in real time? S/he would most likely want to save his/her self above anything. Peppermint's point about simply stopping the vehicle is valid here.
Pepperman wrote:Personally I think that AVs could ultimately be beneficial if the result of moving over to them is the end of private car ownership.
I'd welcome that, though naturally the same (most likely more) money would have to be extracted from former car owners.

We hardly use our vehicle privately anyway - 'saving up' journeys to complete jobs in one go is what we do anyway, though I know we're odd that way. :lol:

Having said that, I use the car a lot for gigs and an AV would be perfect for those (outside rush hours, long boring journeys, tiredness etc).
I experience pleasure and pains, and pursue goals in service of them, so I cannot reasonably deny the right of other sentient agents to do the same - Steven Pinker
Little John
Posts: 8549
Joined: Sat Mar 08, 2008 12:07 am
Location: UK

Post by Little John »

Pepperman, are you saying that there are no circumstances whatsoever in which a driver must make a decision as to which way to act and, having done so, this action may have moral consequences for which they must then morally and legally account for in a court of law?

Because that is the only way what you have said makes any sense.

It is morally completely inconsequential if a computer system carries out the decision of a driver more efficiently than they might have done themselves. But, if that system is making the decisions as well as carrying them out, it is a completely different situation. This cannot be avoided or morally fudged around. Which is what you seem to be doing

EM, comparing the killing of children with congenital illness and the stopping of such children being born in the first place is completely specious. One involves harm to an existing individual. The other simply causes the moral dilemma to not exist in the first place. Unless, that is, one believes it is possible to harm someone who does not exist.

There is, of course, a broader, overarching moral discussion to be had about such selective breeding. But, that is not the same kind of immediate moral dilemma being discussed here.
Last edited by Little John on Wed Mar 01, 2017 2:14 pm, edited 1 time in total.
Pepperman
Posts: 772
Joined: Sun Oct 10, 2010 9:00 am

Post by Pepperman »

Some, but not many. In most cases the response is "slam on the brakes and turn the wheel a bit to try and avoid the worst of it". 99% of drivers simply aren't skilled enough to do much more than that.
Post Reply