Steve Wozniak: AI Will Never Be Smart Enough to Drive a Car

I prefer to listen to today's actual experts in the AI field vs.some guy that retired many years ago on a pile of cash. And this is giving Woz all the kudos he deserves for his time at Apple.
 
Better choice then same body.
lol.

Speaking of which...

I was following this idiot this morning. That's a good 12' ladder nearly vertical on top a F150 pickup bed truck. Do you think AI will be smart enough to avoid this idiot? He came damn close to snagging some lines.

upload_2018-10-1_10-10-14.png
 
Humans blink, can only see accurately in a small area at a time, and sneezing which causes you to have a small spasm and close your eyes. Everyone likes to think they are the most intelligent and best driver on the road and that in no way is there a computer that could beat them. This is just arrogance and insecurity. It really isn't about being smarter or faster than a human, it is about being more aware. Driving at night after a long day, a few too many drinks, anxiety can also mess with driving. I like the idea of a system that can see all around the car, and in a ideal situation, cars are mesh linked to work together. I do enjoy a nice day drive but hjonestly my back and forth to work drive every day I am fine handing that over to the ai while i get an extra hour of sleep lol.
 
What would swerve out of the way of an unexpected danger, an automated car with 100 exabytes of computing power, or a 90 year old grandmother? Never say never.
 
Humans blink, can only see accurately in a small area at a time, and sneezing which causes you to have a small spasm and close your eyes. Everyone likes to think they are the most intelligent and best driver on the road and that in no way is there a computer that could beat them. This is just arrogance and insecurity. It really isn't about being smarter or faster than a human, it is about being more aware. Driving at night after a long day, a few too many drinks, anxiety can also mess with driving. I like the idea of a system that can see all around the car, and in a ideal situation, cars are mesh linked to work together. I do enjoy a nice day drive but hjonestly my back and forth to work drive every day I am fine handing that over to the ai while i get an extra hour of sleep lol.

So you are telling me AI is smart enough to realize the above picture is dangerous?

People really don't understand AI and who make statements about what it can do are well...uninformed

-Signed
-Someone who is teaching AI to others in class.
 
So you are telling me AI is smart enough to realize the above picture is dangerous?

People really don't understand AI and who make statements about what it can do are well...uninformed

-Signed
-Someone who is teaching AI to others in class.

People who have an opinion differently than me are wrong. So is there a point in talking to you?

What is hard for you to understand? It will not be perfect, but if it is 10% less accidents then it would be considered better. We can both create scenarios all day about how humans cant react to one and how ai cant react to the other, it is about improving overall safety not perfection. If perfection is a goal you will ultimately fail.

Also we do use AI as a general term a bit too often, how about for sake of argument we refer to it as software?

Also did that ladder fall and you were able to be far enough back to avoid? Or did nothing happen and neither you nor 'AI' would have had to do anything?
 
I would be interested to find out how many of the autonomous car believers here have worked on real-time, safety critical, high reliability projects like medical devices or avionics. Because in my mind, those are the only opinions that carry much weight. And yes, I do have that kind of experience.
 
This guy is dead wrong. It's already here. A highway populated entirely with Teslas driving themselves at speed would see no accidents.

So when the shitty worn out lane marking confuse it, well there might not be accidents, but they certianly can't be described as "Driving" while they sit there in the middle of the road waiting for a human to take over.

One of the factors AI drivers WILL have to deal with is relying on infrastructure that is AT LEAST as poorly maintained as today.

If people were really shitty at driving, there would be so many more fatalities. Next time you see some asshole endangering everyone, if you can, take note of how many drivers did their job to avoid that shithead's behavior form turning into a disaster.

There are 222 million licensed drivers in the united states. Passenger vehicles log more than 3.2 trillion miles per year in the US. With about 6 million accidents and 37,000 fatalities per year.

Yes, you get annoyed, but for the most part people do their job. You just notice those one or two people who are asshats, and there's always some fo them each and every day. You also ignore the other thousands of people you share the road with that didn't try and kill everyone by being an asshat.
 
So you are telling me AI is smart enough to realize the above picture is dangerous?

People really don't understand AI and who make statements about what it can do are well...uninformed

-Signed
-Someone who is teaching AI to others in class.


It's just regressions. Most of the Musk fanboi's need to start there.
 
This guy is dead wrong. It's already here. A highway populated entirely with Teslas driving themselves at speed would see no accidents.

Says the guy who doesn't have a tesla with autopilot lol.
 
So you are telling me AI is smart enough to realize the above picture is dangerous?

People really don't understand AI and who make statements about what it can do are well...uninformed

-Signed
-Someone who is teaching AI to others in class.
I'm telling you that a human is more likely to miss a thousand other things about themselves that are more dangerous than ladder dumbass. For example, the effect of stress, being tired, having an argument with a spouse, thinking about other tasks you need to do, etc. All of those distract human drivers and make them less effective, and those factors are millions of times more common than encountering the dumbass with a ladder.
 
  • Like
Reactions: DocNo
like this
I'm telling you that a human is more likely to miss a thousand other things about themselves that are more dangerous than ladder dumbass. For example, the effect of stress, being tired, having an argument with a spouse, thinking about other tasks you need to do, etc. All of those distract human drivers and make them less effective, and those factors are millions of times more common than encountering the dumbass with a ladder.

Why should the AI give a shit about some dumb ass hauling a ladder dangerously? You know why it wouldn't because it won't be fucking tailgating the guy like a lot of humans and being too close if something happens. It will be able to brake if something goes bad due to idiocy. I'm all for more testing but that is a stupid example.
 
So you are telling me AI is smart enough to realize the above picture is dangerous?

People really don't understand AI and who make statements about what it can do are well...uninformed

-Signed
-Someone who is teaching AI to others in class.

So are you telling us that people are smart enough to load their vehicles properly and avoid other drivers like this all the time?

And its great that you're teaching this stuff, but condescending comments like that bring absolutely nothing to the discussion.

Why should the AI give a shit about some dumb ass hauling a ladder dangerously? You know why it wouldn't because it won't be fucking tailgating the guy like a lot of humans and being too close if something happens. It will be able to brake if something goes bad due to idiocy. I'm all for more testing but that is a stupid example.

Fucking this. First thing I thought when I saw that photo was "some idiot will be right on his ass."

I've never been in a crash *knock on wood* - but I have enough humility to know that I am no perfect driver. I try to do enough various safety checks to avoid having to make a life or death decision - and in the end it might be pointless because I could get totally blindsided. I don't trust anyone on the road, myself included.

I am also very biased on this stuff - if there is anything that 13 years on southern cali roads has taught me, is that humans are barely in control of their vehicles. I've visited cities all over the country too, usually a different style of crazy but still the same dance.

Except the only way this tech works is if you are all in. Once you train drivers to turn off their brains and let the computer handle it they will not be prepared to take over when it's necessary. And country roads is exactly where you want this tech because highway hypnosis is worse on dark dirt-tracks in the snow. City people shouldn't need to drive at all, they have the bandwidth to go full telecommuting but governments never somehow want to commit to that even though it would be the greenest option. I wonder which vested interest is behind that?

Red herring. Yeah our cities have large scale problems with commuting, and this is the latest approach to do something about it. Telecommuting is no fix-all, nor is auto drive. Public transportation could've been a solution to the traffic problem years ago - but people want the freedom of having their own car and many cities' bus lines are all but useless. We can all agree that it is too easy to get a license - but most people need a car and can demonstrate some capacity to remain in control. We already have different classes of driver licenses, doesn't sound unreasonable to me to change up the system a bit to account for a monitors license vs operators license.
 
Last edited:
  • Like
Reactions: DocNo
like this
I can get my head around self driving cars on the highway and other long distance roads where traffic moves and is relatively patterned/predictable. But put 20,000 AI cars at the football stadium all leaving at the same time and that's where I start to question if the logic can be perfected enough. I don't know if it can be programmed well enough to deal with all the little nuances that occur in tight traffic. Car broke down in the middle of the road, no shoulder available. Traffic lights burnt out, no police directing traffic. There are endless unpredictable scenarios.
 
I can get my head around self driving cars on the highway and other long distance roads where traffic moves and is relatively patterned/predictable. But put 20,000 AI cars at the football stadium all leaving at the same time and that's where I start to question if the logic can be perfected enough. I don't know if it can be programmed well enough to deal with all the little nuances that occur in tight traffic. Car broke down in the middle of the road, no shoulder available. Traffic lights burnt out, no police directing traffic. There are endless unpredictable scenarios.

I think they want to deal with some of what you bring up by using some kind of swarm logic and inter-vehicle communication networks (like ad-hoc). This is further down the road (heh), getting a system that can demonstrate independent control is first priority. There are a lot of issues with this of course, security being the big one. But I think a centrally coordinated traffic system has a lot of potential.
 
So when the shitty worn out lane marking confuse it, well there might not be accidents,

Shitty lane markings confuse me - especially when it rains at night and they all but disappear. For some dumb ass reason Virginia doesn't believe in lane reflectors. And don't give me that snow plow BS - they are all over in Pennsylvania and Ohio; yet somehow they manage to maintain them despite getting WAY more snow than we do.
 
I think they want to deal with some of what you bring up by using some kind of swarm logic and inter-vehicle communication networks (like ad-hoc). This is further down the road (heh), getting a system that can demonstrate independent control is first priority. There are a lot of issues with this of course, security being the big one. But I think a centrally coordinated traffic system has a lot of potential.

Until some hacker "fritzes" the network with bad data. Fords idea of inter car communication is just bad.
 
Until some hacker "fritzes" the network with bad data. Fords idea of inter car communication is just bad.

I thought of this as well. Nothing like an entire network of vehicles disabled due to a glitch or a hacker. Kind of like the entire fleet of Delta airlines being grounded last week due to a central computer problem.

I also don't like the feeling of being a "drone" in my own vehicle. I mean, we're all going to sit there mindlessly, while our self driving cars whiz us along? I don't doubt the technology can be developed. But my skeptical/pessimistic personality thinks this is a bad idea overall.
 
I thought of this as well. Nothing like an entire network of vehicles disabled due to a glitch or a hacker. Kind of like the entire fleet of Delta airlines being grounded last week due to a central computer problem.

I also don't like the feeling of being a "drone" in my own vehicle. I mean, we're all going to sit there mindlessly, while our self driving cars whiz us along? I don't doubt the technology can be developed. But my skeptical/pessimistic personality thinks this is a bad idea overall.


And yet we continure to fly, delta,

My point is people make all sorts of fears turn into reasons we cant do something but the matter of fact is its not about what can go wrong its about how much more right things can be on average. Tomorrow the bridge over the river could collapse, people could die, but if it saves millions of trips per year hundreds of thousands of hours of driving its still worth building a bridge. And yes you might actually have to learn a little to use it, like the bridge could be icy before the rest of the road.
 
It's going to be a very long time before we have all the false positives filtered out of self-driving systems, while still not killing other drivers or pedestrians.

The reason the Uber car hit a pedestrian is because they turned off the more reactive AI algorithms. Because the car was hitting the fucking brakes every five seconds like a timid teenager. That does not go over well when you're driving people around FOR PAY.

The same filtering of "false positives" is why self-driving Teslas keep hitting jersey walls in the middle of a street: because applying emergency brakes or aggressive maneuvers for a lot of "false positives" would turn people off the technology very quickly. So they pretend the tech is already mature, and hope your don't end-up being dead thanks to a (rare, but does happen) false negative.

People will not put-up with self-driving cars that jerk them around constantly, and they will never give the wheel to a car that hits things pretty regularly. just to save the driver's pay. And it's going to be a long time before we get the grown-up AI driver who can actually anticipate road hazards, and slow down at a reasonable pace.

I'd say we're at least 20-40 years out from that quality of AI.
 
Last edited:
I thought of this as well. Nothing like an entire network of vehicles disabled due to a glitch or a hacker. Kind of like the entire fleet of Delta airlines being grounded last week due to a central computer problem.

I also don't like the feeling of being a "drone" in my own vehicle. I mean, we're all going to sit there mindlessly, while our self driving cars whiz us along? I don't doubt the technology can be developed. But my skeptical/pessimistic personality thinks this is a bad idea overall.

they only reason they were grounded was because they couldn't check passengers in, the planes could of flown without any issues. so that's a terrible comparison.

Shitty lane markings confuse me - especially when it rains at night and they all but disappear. For some dumb ass reason Virginia doesn't believe in lane reflectors. And don't give me that snow plow BS - they are all over in Pennsylvania and Ohio; yet somehow they manage to maintain them despite getting WAY more snow than we do.

same problem here.. there's a lot of things that would need to change to make autonomous cars a reality. standardizing how all roads are marked, standardizing lane widths, and standardizing all signs nationwide would be a start but would be expensive as hell and take time given how many roads there are in the world. personally i feel the traffic administrations around the world need to get ahead of the issue long before it becomes a reality to make sure there are specific guidelines that every car manufacture has to follow, not wait til they are here then try to implement it because you damn well know every manufacture will have it's own separate system following it's own rules.
 
Last edited:
"Computers will never be smart enough to beat a chess master"

Chess is a bad example, it's just brute forcing millions of moves ahead and picking the best.


Beating people in real games that have some element of creativity(which is already being done) is much better. Still infantile compared to driving.
 
they only reason they were grounded was because they couldn't check passengers in, the planes could of flown without any issues. so that's a terrible comparison.

Fair enough. But the point is that one small technical snafu took down the entire operation momentarily. Obviously there would need to be redundancy in the design to make it safe and reliable (which can be done).

I admit I do not like change. But I can see the positives of autonomous vehicles. Preventing accidents is at the top of the list in my opinion. Too many bad drivers on the road, under the influence, etc.
 
Chess is a bad example, it's just brute forcing millions of moves ahead and picking the best.

Some experts claimed around the 2000s that "It may be a hundred years before a computer beats humans at Go — maybe even longer", because Go wasn't affordable by brute-force approaches. And it was impossible... until an artificial intelligence did a pair of years ago. I.e. it happened about one order of magnitude faster than the prediction.
 
Some experts claimed around the 2000s that "It may be a hundred years before a computer beats humans at Go — maybe even longer", because Go wasn't affordable by brute-force approaches. And it was impossible... until an artificial intelligence did a pair of years ago. I.e. it happened about one order of magnitude faster than the prediction.

Ummm?

I could have programmed a bot without any AI that would probably dominate at CS GO. What experts are you talking about, any links? It's a bot with perfect aim and 0ms reaction time... derp, you don't need AI to win like that.
 
Welp, in Software Development, that's usually handled with

Else {
};
 
Ummm?

I could have programmed a bot without any AI that would probably dominate at CS GO. What experts are you talking about, any links? It's a bot with perfect aim and 0ms reaction time... derp, you don't need AI to win like that.
Go, not CS GO, lol.
 
Right, dude's been dead for nearly 30 years, you think if he was still alive he would not have kept up? implying that someone that was a pioneer in a field can't figure shit out because he's old? wow. The ghost of Stephen Hawking would like a word with you sir.
There is is the difference though, Stephen Hawking did keep up, he still worked his art. Would Shockley have done the same? Perhaps he would since he came from a different time period than Woz, plus Shockley didn't exactly become a multi-millionaire since the parent company owned the patents for the work. And who knows maybe he would have been the "good company man" and work until retirement, then left on a pension.

But you're probably right though... Shockley would have more insight into the current microprocessors... WAY more than Woz with current technology, since he basically retired from from the field at a young age.
 
Last edited:
Judging by the number of dead deer and damaged cars, not many. But IMO, that is where Tesla and others have failed. Instead of jumping direct to 'Autopilot', why not start with IR sensors that can detect deer and others by the side of the road and project target markers on some kind of HUD and let the human figure it out.
I believe Cadillac had something with IR sensors to detect "warm things" at night and had an overlay on the window... or at least that was their huge tech demo a decade or more back... what happened to that?

But there in lies the big difference, people are still going to be people, we'll still have more or less tunnel vision when we drive we're not changing, and giving us more information to absorb only takes away from our ability to react to what's in front of us. With "AI" you can have sensors in every direction, you can have the ability to detect "warm objects" on the side of the road and if they start moving on an intercept vector for your car the car can do something about it to avoid it, maybe slow, maybe swerve while communicating to other AI vehicles, maybe honking the horn in such a frequency that resembles deer whistles to scare the critters off... based on the the algorithm that detects it's a large hooved creature. Yeah, we're not there today, but the possible ceiling is so much higher than what people can do for someone to dismiss AI as never being as good as a human driver seems quite the stretch.
 
I prefer to listen to today's actual experts in the AI field vs.some guy that retired many years ago on a pile of cash. And this is giving Woz all the kudos he deserves for his time at Apple.
You mean some guy whose's paycheck is dependent on everyone thinking AI is the next big thing vs. some retired millionaire who has no skin in the game, at least financially?
 
You mean some guy whose's paycheck is dependent on everyone thinking AI is the next big thing vs. some retired millionaire who has no skin in the game, at least financially?

Nope, I mean people that have more knowledge than him about AI technology. I thought that was pretty clear but I guess I have to spell it out for those that are less educated.
 
You mean some guy whose's paycheck is dependent on everyone thinking AI is the next big thing vs. some retired millionaire who has no skin in the game, at least financially?
An interesting take on the matter. You could also argue the guy who has no skin in the game also isn't any more of an expert than literally anyone else who reads article online, at which point people are publishing what he's saying (and people listening) as some undeniable "truth" ... why because he's rich?
 
Back
Top