Not for Hawk :)

[Hawk]

I though I had already posted this: http://www.bloomberg.com/news/features/2016-08-18/uber-s-first-self-driving-fleet-arrives-in-pittsburgh-this-month-is06r7on

This month, for public use. It won't be pushed onto reluctant users: you'll be free of not using it if you don't like the idea.

The day it is finally and undoubtedly proven the autonomous vehicles are safer, only motorheads will be against.
Now, you're free to think it will never happen, but what if it happens ? What if it is proven that autonomous cars are safer than human-driven cars ? What will be your position ? I'm curious ...

AI: you've seen too much Terminator ... we're talking about a car that can safely take you from A to B, not about a robot than can self assemble and reason abut how to conquest the galaxy.

Yes, I saw that, but that is with a supervised human driver - That's hardly driverless vehicles is it, and as I stated when commenting on it in one of my posts above, it just shows they are not 100% confident that driverless vehicles will work as they should without making a mistake. Just proves they are not really ready yet.

If it proves itself a lot safer than human drivers then I'll support it.... Don't get me wrong, it has it's potential but I just don't think it is yet ready to be trusted in public.... I think there will be tragic accidents that will involve not just other cars and drivers but also pedestrians on the sidewalks that probably wouldn't have happened if a human driver had been in control. That will be the tragedy of debugging these vehicles in public use. It will only take one tragic incident to destroy public confidence in these vehicles, so I hope they've got them very well debugged before there release?

True AI is a self aware artificial entity. That's why I did classify it as "True AI", your probably talking about computers with pre-programmed neural networks that take a situation and compare it's results with say two other systems and choose the majority decision; that's not true AI although they do appear quite intelligent for the use they're designed for...... Though I doubt very much they are using AI for these vehicles decision making as it is still too unreliable I would've thought. 

Hawk.

[HornetMaX]

Yes, I saw that, but that is with a supervised human driver - That's hardly driverless vehicles is it, and as I stated when commenting on it in one of my posts above, it just shows they are not 100% confident that driverless vehicles will work as they should without making a mistake. Just proves they are not really ready yet.

There are multiple reasons for that: situations that requires human to re-take control (note, these are mostly not emergency situations, but stuff like roadworks for example), reassuring people that are completely scared by that, ... The end goal is self driving, there's little doubt.

If it proves itself a lot safer than human drivers then I'll support it.... Don't get me wrong, it has it's potential but I just don't think it is yet ready to be trusted in public.... I think there will be tragic accidents that will involve not just other cars and drivers but also pedestrians on the sidewalks that probably wouldn't have happened if a human driver had been in control. That will be the tragedy of debugging these vehicles in public use. It will only take one tragic incident to destroy public confidence in these vehicles, so I hope they've got them very well debugged before there release?

This is the common fallacy. People will get absolutely mad about the first big crash of a self-driving car, especially if the crash is one that somebody would come up and label as "a crash a human could have avoided". Independently if the statement is true or false (that a human would have avoided it), what the people forget is that for this crash that a human could have (eventually) avoided there have been 10 others that the self-driving car *has* avoided and the human would have not avoided.

That's the key thing: they don't have to make it perfect, they have to make it better than human drivers. That would be enough.

True AI is a self aware artificial entity. That's why I did classify it as "True AI", your probably talking about computers with pre-programmed neural networks that take a situation and compare it's results with say two other systems and choose the majority decision; that's not true AI although they do appear quite intelligent for the use they're designed for...... Though I doubt very much they are using AI for these vehicles decision making as it is still too unreliable I would've thought. 

Depends what you call AI. Self-awareness is a bit too extreme for me as a necessary condition. But it doesn't matter if you call it "AI-driven car" or "self-driving car with a dumb fixed algorithm on it": if it saves lives, it's progress. And I'm in.

[Hawk]

I agree that if it saves lives then that is a good thing, but I just think they have done a very poor job of promoting that if it is true, as well as a poor job in showing the public how safe they are. That breeds suspicion in the general public that they are hiding the possible truth that in reality they are not truly as safe as a human at avoiding unforeseen incidents that happen everyday on the road. And as I say, the fact that they feel the need to have a real human supervisor shows that they know these driverless vehicles are not safe to be totally driverless, and if a common obstacle like road-works mess up there ability to drive safely, then to be honest, that doesn't bode well for their future safety record does it?
I would like to know how it handles pulling out to overtake an object in the road if another vehicle already overtaking a whole line of cars is rapidly coming up at a distance from behind? Can the driverless car see and judge the speed of that car in the opposite lane coming from behind when there maybe another car close up to it's rear bumper? Or will it not sense it and pull out into the overtake lane and cause the fast overtaking car from behind to have to violently brake to avoid hitting you? Or at worst cause an accident by doing so? These are the sort of basic tests they need to show the public how these driverless cars handle these potentially dangerous situations.... Question is: Why haven't they? Is it pressure from investors to get the product out and making money before it's actually fully safe and ready to be released onto the public roads? That would not be a first would it.... In this society money talks, money is our God, and unfortunately peoples  lives come second at best in comparison to these sort of people..... Any current problems that cause accidents will just be put down to teething problems and fixed at a later date when the money has already started rolling in to the investors..... That's a classic product pressure release deadline situation we see all the time with major software developers who rely on major investors.... Ring a bell? 

But I agree that if driverless vehicles do turn out to be safer and more efficient at getting you from A-B on the road then that is a good thing, but convincing the public to give up their total driving independence to a computer will be a very real big mountain to climb, and I personally think the only way to achieve it will be for the government to ban self drive cars from being sold after a certain date/year. But it would be a very brave government to do something like that. Lol 

Hawk.

[HornetMaX]

I agree that if it saves lives then that is a good thing, but I just think they have done a very poor job of promoting that if it is true, as well as a poor job in showing the public how safe they are.

All the info is out there. Unmanned vehicles are under strict scrutiny, for obvious reasons.
Don't worry, the day somebody starts to sell something, they will let you know how safe it is.

I would like to know how it handles pulling out to overtake an object in the road if another vehicle already overtaking a whole line of cars is rapidly coming up at a distance from behind? Can the driverless car see and judge the speed of that car in the opposite lane coming from behind when there maybe another car close up to it's rear bumper? Or will it not sense it and pull out into the overtake lane and cause the fast overtaking car from behind to have to violently brake to avoid hitting you? Or at worst cause an accident by doing so?

You really think that the legions of researchers and engineers that are currently working on this have not thought about such a simple test ?

The cars are being tested on the roads because they are past the point of simple "what-if" tests; they need to go to the road because that is the only thing that would prove if they are (or not) better than humans at driving safely.

[TFC]

Hell I'd just be happy to follow a vehicle that actually joins a motorway / dual carriageway at the right speed.

People who drive at 40 down a slip road, join, then speed up to 80 once on the road need to loose their licence.



Oh, and those who drive at 45 on 60 roads then don't slow down and carry on doing 45 through 30mph villages.

Oh and those who use the outside lane of a roubderbout to take the 4th exit..

In fact I'd be more concerned about self drive cars coming up against people who don't use the road properly, as it may 'not compute'

Seriously though I'm full of road rage, kept to myself and whoever is in the car with me of course

[Hawk]

Indeed! 
Those are questions that need answering, and if these driverless cars are capable of handling such situations then why don't they promote that for their safety and show us all just how good and safe these driverless cars are in those varied situations and circumstances.

An old saying comes to mind: "Trust is earned, not given". 

Hawk.

[HornetMaX]

Trust is being earned by trialing the self-driving cars on the road.
Testing them in a controlled environment will prove little.

What do you do with a human willing to learn how to drive ? You teach him and then ... you send him out on the public road (with a supervisor for a while).