What are the current limits of artificial intelligence? & their weaknesses

[Zzzz1]

Tinfoil anecdotes. When that happens you should worry about other things than traffic jams...

Food for thought!
Wait until someone figure out how to hack the GPS satellite and then do ur statistics!
By the way the Russians and the Chinese already have the tools to do just that. And just imagine what will happen to Cali freeways if GPS satellites get blown out of the sky.

 

[SteveM]

The stats speak for themselves. Self driven cars TODAY are already statistically speaking 100 times safer than an average human driver.

I emphatically disagree with this. Safety results generated from a test where one car drives around San Francisco at 20 mph while being monitored in the real-time by 10 engineers, is a far different proposition than 120,000 automated cars/tractor trailers moving at 60 mph during rush hour traffic in Washington DC while it is raining out.

 

[Zzzz1]

That and way beyond have been tests already completed years go. There are optimizations that engineers work on that deal with edge cases, situations any human being, even expert professional drivers find mind boggling and would not handle in the best possible way under stress. You should really read up on the latest tests and advancements. Fact remains that human error that leads to accidents on the road is several orders of magnitude higher than "errors in judgement" by algorithms inside self driven cars.

... is a far different proposition than 120,000 automated cars/tractor trailers moving at 60 mph during rush hour traffic in Washington DC while it is raining out.

 

[Turveyd]

That and way beyond have been tests already completed years go. There are optimizations that engineers work on that deal with edge cases, situations any human being, even expert professional drivers find mind boggling and would not handle in the best possible way under stress. You should really read up on the latest tests and advancements. Fact remains that human error that leads to accidents on the road is several orders of magnitude higher than "errors in judgement" by algorithms inside self driven cars.

When we stop all human driven cars let the cars all talk to each other and plan when to cross a junction without having to stop and look at all, then maybe.

Ofcourse I'd have much more faith if they get the simple things ie Windows or Android OS working this decade.

 

[Zzzz1]

Well you can of course deny facts and simple statistical arguments, that is your choice. And you can pepper your argument with a little joke on the side relating to mobile phones (which by the way are neither perfect but handle the job way better than stringing together two opened baked bean cans and trying to make conversation). That does still not change the current facts of where AI in self driven cars is and which edge cases engineers are looking to handle at the moment. They are done and finished with the AI making decisions in Any Weather conditions or road conditions. They are currently working on how to interpret and react to behavior of other human drivers on the road who make blatantly irrational and stupid decisions. Those are optimizations that go way beyond the quality of reaction by even expert human drivers.

I will retreat from this discussion unless we can all at least be up to date on where AI currently stands. It's kind of pointless to argue with someone who believes the status quo is on the level from many years ago and that it is not sufficient to meet real life situations.

When we stop all human driven cars let the cars all talk to each other and plan when to cross a junction without having to stop and look at all, then maybe.

Ofcourse I'd have much more faith if they get the simple things ie Windows or Android OS working this decade.

 

[Turveyd]

Well you can of course deny facts and simple statistical arguments, that is your choice. And you can pepper your argument with a little joke on the side relating to mobile phones (which by the way are neither perfect but handle the job way better than stringing together two opened baked bean cans and trying to make conversation). That does still not change the current facts of where AI in self driven cars is and which edge cases engineers are looking to handle at the moment. They are done and finished with the AI making decisions in Any Weather conditions or road conditions. They are currently working on how to interpret and react to behavior of other human drivers on the road who make blatantly irrational and stupid decisions. Those are optimizations that go way beyond the quality of reaction by even expert human drivers.

I will retreat from this discussion unless we can all at least be up to date on where AI currently stands. It's kind of pointless to argue with someone who believes the status quo is on the level from many years ago and that it is not sufficient to meet real life situations.

I'd be happy current level of Tech to see Computer cars driving around my local town / streets some of the useless drivers around sounds good, maybe even drive me at low speeds 30mph and below as modern cars unlikely to be injured at those speeds so not a great risk, Even happy on Motorways as all cars going the same way and relatively simple to do.

A and B roads and higher speeds, I'd still prefer to be driving for a few generations.

Computers interrupting visual and radar data is tricky, tiniest of errors at 60mph on a A Road and your in a tree or worse hitting on coming traffic.

It's still not AI though, IQ of the google car is still Zero.ZeroZeroZeroZeroZeroZero, it's just code and logic based on all the sensors.

 

[SteveM]

Fact remains that human error that leads to accidents on the road is several orders of magnitude higher than "errors in judgement" by algorithms inside self driven cars.

How can you possibly propose that when there has been zero widespread implementation of driverless technology?

This is the equivalent of saying "pilot-free 747 passenger aircraft have a far better safety record than 747 aircraft flown by pilots." Data sample is non-existent.

 

[Zzzz1]

Driverless cars habe driven around for many years. You just don't see them because they were all on experimental stage and only recently regulators are considering widespread implementation. I never said we should let unproven technologies lose on human kind but the amount of misinformation in this thread is simply mind boggling and outright shocking. Clearly written by people who have no fuxxing clue about this technology. And written by people who freak about 1 minor accident (most incidences with self driven cars involved are not even caused by such cars but by other human drivers who drove carelessly) by a self driven or ai assisted car when during the same weekend dozens died in car accidents because of stupid human errors. One is overreported the others nowhere mentioned in your USA Today. Brainwashed people.

How can you possibly propose that when there has been zero widespread implementation of driverless technology?

This is the equivalent of saying "pilot-free 747 passenger aircraft have a far better safety record than 747 aircraft flown by pilots." Data sample is non-existent.

 

[Turveyd]

Driverless cars habe driven around for many years. You just don't see them because they were all on experimental stage and only recently regulators are considering widespread implementation. I never said we should let unproven technologies lose on human kind but the amount of misinformation in this thread is simply mind boggling and outright shocking. Clearly written by people who have no fuxxing clue about this technology. And written by people who freak about 1 minor accident (most incidences with self driven cars involved are not even caused by such cars but by other human drivers who drove carelessly) by a self driven or ai assisted car when during the same weekend dozens died in car accidents because of stupid human errors. One is overreported the others nowhere mentioned in your USA Today. Brainwashed people.

I think you should go watch all of the Terminator films, including Salvation you brought that 1 upon yourself.

We haven't got access yet to review the technology it's still in R&D Phase, I think or was Uber using it live recently ??

Yes Humans make mistakes, Computers well coders do aswell, hopefully it'll get to a stage where it's 10x's safer to be driven by a computer than a person, Tesla's autodrive is already doing great, but it's not 100% computer driven yet.

Chill your taking this wayyyy to personally!

 

[Zzzz1]

I don't take it personally. I just have very short attention span when it comes to misinformation. Fact is the technology is mature enough to be implemented in small stages. We have driving assisted ai in production now and next up is completely autonomous self driven cars by AI. Both are already today way safer than human drivers yet regulators don't want to lean too far out of the window for fear of reprisals. It's the same ridiculous mindset as someone who freaks over a death that in whatever capacity involved marijuana and now wants to ban marijuana but can sleep well ignoring the fact thousands die each year because of alcohol and tobacco abuse. We live in a funny world, inhabited by mostly average iq people who report our news and regulate our everyday life.

You don't find it ridiculous how a single engineer made news about how he was paid 120 million usd by Google yet we don't celebrate the absolute breakthrough in technology that we are near completion to have self driven cars at our disposal that reduce deaths on roads by like 99.5%, once fully implemented?

I think you should go watch all of the Terminator films, including Salvation you brought that 1 upon yourself.

We haven't got access yet to review the technology it's still in R&D Phase, I think or was Uber using it live recently ??

Yes Humans make mistakes, Computers well coders do aswell, hopefully it'll get to a stage where it's 10x's safer to be driven by a computer than a person, Tesla's autodrive is already doing great, but it's not 100% computer driven yet.

Chill your taking this wayyyy to personally!