Mustang v. Plaid

[MAGS1]

You know, you say that in a joking way, but you're not far off. The fact is, nobody has the slightest clue how our minds work, so replicating them with a computer is impossible from that fact alone. And the more we do understand, the less it seems like a computer can do what our brains do.

Simply put, unlike what has been assumed for a few decades, our brains aren't just computers running software. There's something else going on that can't be mathematically modeled in a way that computer algorithms can duplicate.

Honestly, it’s what makes humans be humans. No 2 people think or act exactly the same. I would fear that AI replication would take a lot of that uniqueness away.

Sponsored

 

[wynand32]

Honestly, it’s what makes humans be humans. No 2 people think or act exactly the same. I would fear that AI replication would take a lot of that uniqueness away.

I don't think that's anything we'll have to worry about in our lifetimes (and that means anyone alive today, no matter how old). You'll need to worry if you have a menial job that can be easily replicated, but that's about it for a long time.

 

[HoosierDaddy]

I don't think that's anything we'll have to worry about in our lifetimes (and that means anyone alive today, no matter how old). You'll need to worry if you have a menial job that can be easily replicated, but that's about it for a long time.

People with any such job should unite and copyright. Then they can live off the royalties AI will have to pay to do their job.

 

[andrewtac]

And yet it can't even do its simple job properly. Let alone work continously in all imaginable situations.

I have had the assisted brake work for me in situations it should have. I realize we are disagreeing on completely different things now. I was suggesting it is possible to have a solution that will mostly work, you are looking for something 100%. We should have started here, I agree there will never be anything that works in all imaginable situations, you are correct current tech can't do that and likely will never be able to as long as we are designing it.

 

[wynand32]

I have had the assisted brake work for me in situations it should have. I realize we are disagreeing on completely different things now. I was suggesting it is possible to have a solution that will mostly work, you are looking for something 100%. We should have started here, I agree there will never be anything that works in all imaginable situations, you are correct current tech can't do that and likely will never be able to as long as we are designing it.

And the thing is, a self-driving system needs to be able to work in "all imaginable situations" to be viable. That's what distinguishes the human mind from AI, the latter of which is mathematically/algorithmically designed, thus entirely deterministic and necessarily based on human-limited programming. They hope to "fix" that problem through machine learning and neural networks, and by feeding a steady stream of data into the system. But the system will never be able to actually "think" and thus handle situations that aren't programmed in.

We may not have the pure brute-force "reflexes" that a computer can muster, but we can adapt on the fly to novel situations. Clearly, we're not perfect, hence so many accidents caused by human error. But we're generally able to do what computers cannot, which when put in the same situations will fail miserably.

To get back on point, Tesla sells a $15K "self-driving" package that isn't nearly ready to act autonomously and probably never will be. IMO, it's straight-up fraud and if I had to predict, I'll bet there's a class action lawsuit sooner rather than later. Remove the disclaimer about keeping the driver involved and that class action suit will shut the company down.

 

[andrewtac]

And the thing is, a self-driving system needs to be able to work in "all imaginable situations" to be viable. That's what distinguishes the human mind from AI, the latter of which is mathematically/algorithmically designed, thus entirely deterministic and necessarily based on human-limited programming.

We may not have the pure brute-force "reflexes" that a computer can muster, but we can adapt on the fly to novel situations. Clearly, we're not perfect, hence so many accidents caused by human error. But we're generally able to do what computers cannot, which when put in the same situations will fail miserably.

To get back on point, Tesla sells a $15K "self-driving" package that isn't nearly ready to act autonomously and probably never will be. IMO, it's straight-up fraud and if I had to predict, I'll bet there's a class action lawsuit sooner rather than later. Remove the disclaimer about keeping the driver involved and that class action suit will shut the company down.

Same here as well, we should have started with this. I agree with you, there is absolutely nothing that will work in all situations. It will never work 100%, ever. I did not realize we too were talking about different things.

Humans are not capable of reacting as fast, sorry this is inaccurate. Look at any test project for a vehicle (air, land, sea). There is always human reaction time, that is now orders of magnitude longer than what something automated can do.

 

[wynand32]

Same here as well, we should have started with this. I agree with you, there is absolutely nothing that will work in all situations. It will never work 100%, ever. I did not realize we too were talking about different things.

Humans are not capable of reacting as fast, sorry this is inaccurate. Look at any test project for a vehicle (air, land, sea). There is always human reaction time, that is now orders of magnitude longer than what something automated can do.

I won't argue the point about human reflexes, I don't doubt what you're saying is true. If I'm wrong about some specific systems being faster, like assisted braking, then that's fine. It just makes me believe even more strongly that Elon Musk is a complete fraud.

 

[Phils15GT]

My Plaid has FSD capability and I rarely use it. Only long and straight drives do I use it. The FSD has a long way to go. Way too finicky, slow turning with random breaking. Definitely don't trust it.

 

[wynand32]

My Plaid has FSD capability and I rarely use it. Only long and straight drives do I use it. The FSD has a long way to go. Way too finicky, slow turning with random breaking. Definitely don't trust it.

So you wouldn't use it in downtown NYC driving?

 

[andrewtac]

I won't argue the point about human reflexes, I don't doubt what you're saying is true. If I'm wrong about some specific systems being faster, like assisted braking, then that's fine. It just makes me believe even more strongly that Elon Musk is a complete fraud.

My experience had been with aircraft and control systems/sensors for them. I am willing to bet a large sum of money that a pilot will not depart an F35B in jetborne flight due to pilot error, while almost all Harrier departures in jetborne flight were due to this. Not reacting properly and fast enough. The navy is cutting large portions of training on the f35c, as the automated flying is far superior coming aboard the boat. The current stuff on cars is very basic compared to this stuff. Again none of it is cost effective or would allow for a profit unless the customer was the US govt.

 

[wynand32]

My experience had been with aircraft and control systems/sensors for them. I am willing to bet a large sum of money that a pilot will not depart an F35B in jetborne flight due to pilot error, while almost all Harrier departures in jetborne flight were due to this. Not reacting properly and fast enough. The navy is cutting large portions of training on the f35c, as the automated flying is far superior coming aboard the boat. The current stuff on cars is very basic compared to this stuff. Again none of it is cost effective or would allow for a profit unless the customer was the US govt.

I don't know much about this particular application, but I'm guessing that it's a much more controlled and predictable environment. That's not to say that the systems don't face challenges, just not the same as with driving a car on crowded streets.

 

[Mike Pfeifer]

I don't know much about this particular application, but I'm guessing that it's a much more controlled and predictable environment. That's not to say that the systems don't face challenges, just not the same as with driving a car on crowded streets.

It probably doesn't have to take into consideration a random plane that might just pop in the way while landing, or a kid jumping out from behind a plane, etc, etc... But, of course, the challenges are very much still there - you don't want X billion dollar planes crashing all the time because it couldn't take into account all the info it needs. I'm sure there is also a failsafe, where if it's not 100% certain due to sudden cross winds or whatever, the pilot has to take over. And a military pilot will absolutely take over, unlike jimmy from youtube driving from the back seat while showing off the full self driving.

 

[wynand32]

It probably doesn't have to take into consideration a random plane that might just pop in the way while landing, or a kid jumping out from behind a plane, etc, etc... But, of course, the challenges are very much still there - you don't want X billion dollar planes crashing all the time because it couldn't take into account all the info it needs. I'm sure there is also a failsafe, where if it's not 100% certain due to sudden cross winds or whatever, the pilot has to take over. And a military pilot will absolutely take over, unlike jimmy from youtube driving from the back seat while showing off the full self driving.

Exactly. Impressive technology for sure, but not really what we're talking about here.

 

[Bulldog9]

Arguing whether a Tesla is or isn't a better racer than a Mustang is like arguing whether Angela Merkel cooks and irons shirts better or faster than Charlize Theron. Who cares? If given a choice, it's still Charlize that most normal men would pick, even if it meant they'd have to order pizza and to wear wrinkled shirts for the rest of their lives.

I see nothing in this post other than Charlize Theron..................

 

[Vlad Soare]

Same here as well, we should have started with this. I agree with you, there is absolutely nothing that will work in all situations. It will never work 100%, ever. I did not realize we too were talking about different things.

We were talking about self-driving cars. Now, our definitions of self-driving may be different. To me, that means a car in which I can sit in the back and read a newspaper while it takes me to my destination, just like I do in a taxi.
I don't think this can be done with the current technology, no matter how much money you throw at it.
If a car can do 99% of my tasks, while still requiring me to watch the road and be ready to intervine for the remaining 1%, then I don't consider that self-driving.

There are many situations when you need to establish eye contact with another driver, to tacitly state your intentions and understand theirs. A machine isn't able to do that and never will be.

Humans are not capable of reacting as fast, sorry this is inaccurate. Look at any test project for a vehicle (air, land, sea). There is always human reaction time, that is now orders of magnitude longer than what something automated can do.

True, but humans can do more than reacting. They can think ahead.

The fallacy is to think that I, as a driver, just sit there waiting for something to happen, and then react to whatever has happened. That couldn't be further from the truth. Of course a machine will apply the brakes faster than I could when a child runs in front of the car. I agree with that. However, in real life I will be aware of that child, and of their likelihood of running in front of my car, long before they actually do it. When they do it I will be prepared. I'll have already started slowing down long before that point, and will be ready to slam on the brakes if necessary before they even get directly in front of the car (which is the moment when the automatic system would trigger). In fact, I may not even need to slam on them at all. I may be able to stop smoothly, due to having started a lot earlier.
It's amazing how many things around me I am aware of when I'm driving, things that trigger countless reactions and preparations before anything actually happens. There aren't enough sensors and radars in the world to see what I'm seeing and to do the mental preparations that I'm constantly doing, most of them without even realizing. I know, for instance, that a driver is extremely likely to pull out in front of me long before he does it, just by analyzing his body language and many other details. I'm not even consciously aware of all of those details; my brain does it automatically. I can't tell you how I knew he would pull out. I just knew. I felt it.

Sure, if we compare these automated systems to people who simply sit in the driver's seat, looking straight at the road ahead, oblivious to their surroundings, just waiting for something to happen and to react to, then they are (or at least have the potential to be) much better indeed. But those aren't the drivers I'm talking about. Whoever drives like that shouldn't have got a driver's licence in the first place. I'm talking about machines that can replace proper drivers, not stupid vegetables that just happen to possess a driving licence.

Sponsored