A glimpse of the future?

[HughBanton]

OK playmates, probably time to get serious again. (Otherwise Spogg will shut us down soon, and give us one of his damn good thrashings !)

I've actually always been a massive fan of self-driving vehicles, in principle, and I can see a very changed world once it has comes to pass, as it undoubtedly will in time. I can imagine it changing the world as much as the internet has .. for better or for worse. Not just cars, but air-flight (oo eck), trains, sea travel, and especially drone deliveries - big time. (Plus here's "take us to the country pub Otto, and collect us later .. when I ring". If there's any country pubs left that is ...)

So I wish Malc every success in his latest venture.

H

[deraudrl]

I've actually always been a massive fan of self-driving vehicles, in principle, ...

So I wish Malc every success in his latest venture.

Yeah, it's not like there's anyone with qualifications and resources already working on this problem. Oh wait...

[Spogg]

OK playmates, probably time to get serious again. (Otherwise Spogg will shut us down soon, and give us one of his damn good thrashings !)

And don’t forget I know where you live so I could apply it in person.

Seriously, I don’t think the term "Artificial Intelligence" is an appropriate name, because I don’t see any actual intelligence. What we call AI is really just fantastically complex and clever database processing. There is no agency like we as humans have. It’s a simulation of intelligence and Alexa doesn’t understand the word “cabbage” like we do.

For self-driving vehicles the database would have to take account of every possible thing that can happen, in all conditions and in any order, and very fast too. This in turn means that the designers of the algorithms and those who train the neural networks need to be all-seeing. Remember that plane that crashed because it “thought” it knew better than the pilot? Of course they say it’s fixed now, but I fear that only the situation that arose is fixed.

So I’m far less optimistic than many that we’ll get safe self-driving cars at any point. I do believe we can use "AI" systems to provide useful driver assistance however, but I would always want a big BYPASS switch.

[adamszabo]

Hmm, I have a slight feeling that we are somewhat "mocking" their project about the self driving cars. I am sure they are working hard on their ideas, even if other companies might do the same. Competition is always a driving force to make better products no? In the same vain I could ask, why are any of us even bothering to make vsts, there are hundreds out there? Its because we have our own ideas and vision of what we want to achieve just like they have their own vision of self driving cars. If they can make it more safer and reliable then I wish them success!

[deraudrl]

So I’m far less optimistic than many that we’ll get safe self-driving cars at any point.

Your definition of "safe" appears to imply the need for perfection of operation in every conceivable situation. That is something that human drivers clearly lack, as demonstrated in deadly fashion on a daily basis. Self-driving cars don't need to be perfectly safe, they just need to be safer than human-driven cars. On average, that's a rather low bar.

[deraudrl]

Hmm, I have a slight feeling that we are somewhat "mocking" their project about the self driving cars.

Apparently I was too subtle in my mocking...I'll try to do better in the future.

They're the better part of a decade late to the party and several orders-of-magnitude light on resources. It's a bit like trying to compete with Airbus by marketing DIY Wright Flyer replica kits.

[trogluddite]

Irrational humans are the problem, really. If you could have a road system which was used ONLY by self-driving cars, the technology could be orders of magnitude simpler. The problem is that they'll be sharing the roads with human drivers, cyclists, pedestrians, etc., all of whom can behave extremely unpredictably, or even maliciously. Milton Keynes, where my family are from, is a formally planned "new town" where motor traffic is isolated from cylists and pedestrians in much of the city, and I've seen the difference which even that can make.

However, the biggest problem may not actually be the technology at all - there needs to be agreement on how liability is resolved if and when there are accidents, and there seems to be very little action as yet by legislators or the insurance industry in this respect. Rightly or wrongly, people DO expect technology to be perfect, so the potential for some extreme legal wrangling is very high once self-driving cars are let loose on public roads. I agree with @deraudrl that it's very likely that they'll eventually be safer than human drivers by any rational measure, but I don't see that argument washing with the majority of citizens who end up getting unlucky.

Having said all that, if it could happen, I love the idea of self-driving cars in principle. My neurological disabilities mean that I'm simply unable to drive, and current public transit, taxis, etc. are impractical for so many things (going on a touring holiday, for example, or even commuting to jobs that I might otherwise be suitable for). Being able to simply rent the transport that I really need as and when I need it could massively improve quality of life for many, many people.

Hmm, I have a slight feeling that we are somewhat "mocking" their project about the self driving cars

Yes, I think that's a fair observation. But I also think it's fair to be skeptical of hyperbole - getting up unrealistic hopes could actually be damaging in the long run, as not meeting people's expectations often leads to backlashes, or at least, damaged trust. Just as competition can drive improvement, so can criticism, humorous or otherwise.

[Spogg]

Don’t get me wrong; I would love to see self-driving cars being a success, especially for those like trog who, for whatever reason, are excluded from the freedoms that being able to drive brings.

Also I can see great benefits in the spin-offs for technology in general, like we got from the space race last century. It could be that Cavonix will make advances that ultimately benefit FlowStone in some way and of course I welcome that.

I think part of my scepticism is based on my feeling a potential lack of trust in autonomous systems, and that’s based not only on PCs and applications doing weird and unexpected things but also on my previous job servicing sophisticated medical X-ray equipment. The tiniest component or line of code can sometimes wreak havoc if it fails.

Another thing that worries me is the quality of engineering that would go into these cars. Can manufacturers afford to have built-in redundancy and dynamic failure analysis, or will such things be precluded due to unit costs? In terms of reliability aviation comes first, followed by medical and finally domestic. It might surprise you that medical equipment reliability is rated in the middle. This is because patients rarely die due to technology failure, but a plane coming down is another thing. However, a rogue car could also take many lives in a bad situation, but to apply aviation standards to a car might make it too expensive to be commercial, so I worry that appropriate standards won’t be met.

[tulamide]

Not to disappoint anyone, but this is a joke. Self-driving cars is more than an app with a map. Car manufacturers invest billions in the development, this "startup" can't even keep up with the speed of development, let alone build a car and fit it with AI.

The biggest hurdle so far is not what people might think. It's ethics. Yep, the really tough questions of what to do in really dangerous situations. Your car becomes unstable. If you stir left you will hit a woman with her baby. The chance they will be hurt is 100%, the chance they'll die is 50%. If you stir right you will hit an old man. The chance he dies is 100%. What to do?

Truth is, we don't make decisions in such situations, where only the blink of an eye is between you and a catastrophe. Instead we act instinctively, and that involves life experience. AI can't act instinctively. It has to make a decision. And that's the ethic problem. Is it ok to kill an old man? He won't live very long anyway, right? Or is it acceptible to hit the woman with her child, as there is a chance they'll survive? And what about the passengers in the car? Should the car decide based on the well-being of them?

Everyone has an opinion on such a scenario. And all of them will be valid. None of them are machine decisions though.

First steps have been made installing a general set of rules regarding ethics in split-second decisions. Suggestion is that the car should always avoid hurting humans and should never differentiate between age, gender, race or amount of humans. And that makes above situation precarious. No matter the decision, the car would break this rule. But it can't. So what's the outcome?

There is much more to it. Here's some material about it.
https://arxiv.org/pdf/1802.04103.pdf
https://link.springer.com/article/10.10 ... 016-9745-2
https://www.sciencefocus.com/future-tec ... -kill-you/

[trogluddite]

@tulamide
Excellent articles, thanks for linking those. The Nyholm/Smids paper in particular I thoroughly enjoyed reading - a really solid and clear explanation of the severe limitations of "trolley problems" when thinking about real-life cases (perfectly fills in the factual side of what the SMBC cartoons I posted were trying to get at, I think).