Artificial Intelligence: Into the Real World

[John F]

This week, the cable channel NHK World is running a program in its Documentary series titled "AI: Into the Real World." I recommend you see it if you can. It shows that AI is now being used for more consequential purposes than beating the world champion at chess, go, or whatever game. A Tokyo taxicab company has installed AI in its cabs to guide drivers to areas where they are most likely to pick up fares. That benefits both the drivers and people wanting a cab - no problem. But a criminal court in California is using AI to determine the length of prison sentences and whether to grant parole. The judge supposedly makes that decision, but in the one case shown in the documentary, she does what the AI recommends. And in South Korea, the legislature has hired an AI programmer to create an AI politician which presumably would be immune to corruption and the influence of lobbyists.

In all these cases, the final decision is up to a human being, or a collection of them. Korea's AI politician would have no vote in the legislature. But in the documentary it's clear that the human decision-makers, whether a taxi driver or a judge, basically does what AI tells her. It's easier than exercising one's own judgment and taking personal responsibility. And if human politicians are judged by whether they conform to the recommendations of the robot, then the voters are letting the robot do their thinking for them, and the robot is in effect governing the country.

Also, AI provides answers to the question it's designed to deal with, in many cases answers that can be validated in the real world, but it does not reveal how it arrives at its answers. In a game like chess, it's possible to figure out the logic behind the AI robot's play, but in criminal court? Not only does the judge not know the robot's reasoning behind the sentence she is passing, but the AI programmer himself often doesn't know how his own creation arrived at its conclusion.

[jbuck919]

I've read a great deal about this. It would be all right if a way were found to administer AI for the general benefit of the population, but that's not what is going to happen, is it? Instead, the extremely wealthy will find a way to use it to take all the advantage for themselves and leave the general population in penury. You and I are just lucky that we're old enough not to live to see the mortal nature of this outcome.

[John F]

I don't think so. AI has possible applications in almost everything. You and I don't drive cars but most Americans do, and many now have AI built in to prevent collisions etc. And there are many possible medical applications and some already in use; one of the NHK World documentaries says that AI enables extremely precise cancer detection. In effect, just about anything that requires natural intelligence may be a candidate for some form of artificial intelligence, for better or worse.

[jbuck919]

I don't think so. AI has possible applications in almost everything. You and I don't drive cars but most Americans do, and many now have AI built in to prevent collisions etc. And there are many possible medical applications and some already in use; one of the NHK World documentaries says that AI enables extremely precise cancer detection. In effect, just about anything that requires natural intelligence may be a candidate for some form of artificial intelligence, for better or worse.

If you were addressing me, I've been driving since I was 16 and still drive every day. I don't have the benefit of a great metropolitan transit system. The fact that cars are infinitely better than they were when I was young is irrelevant. The material you refer to is not talking about that kind of AI. Instead, it involves replacement of humans, which is not necessarily an evil thing, but can be exploited that way, and probably will be by the unscrupulous forces that now command our society.

[John F]

Sorry! I got you mixed up with someone else. But I think you're mistaken about AI. The system used by Tokyo taxi drivers does not replace humans. Neither does some of the AI being developed for medical use; it's a diagnostic tool, like x-rays, to help doctors do better for their patients. Some people may lose their jobs, as many have when robots replaced them on factory assembly lines. But those machines are not capable of thinking for themselves, which is the object of artificial intelligence - that's why it's "intelligence."

As for cars, an increasing number now brake themselves when in danger of hitting someone or something, and sound an alarm when the car wanders over the lane markings without signaling; the next step in this development, and prototypes already exist, is cars that drive themselves. This may not appeal in the U.S., where despite the frightening statistics we believe we're better drivers than we actually are. But in rural Japan, where the population is aging and older drivers accidentally hit the accelerator instead of the brake pedal, a self-driving car may be the only alternative to limited or no transportation at all.

[absinthe]

A subject that interests me. It's been my belief for a while that AI "machines" are the next evolutionary step. It's human arrogance to think of itself as the end of the evolutionary line. So I was intrigued to find Max Keiser reporting his belief that, the way things are progressing, AI will eventually make the human redundant. Much has been debated about "singularity" but as machines are developed that can improve themselves so it'll take off. Advances are accelerating. There are projects on the go now attempting to decode the mind. A few reports have appeared. Just one:

https://www.asianscientist.com/2017/06/ ... agination/

[absinthe]

Change the world for the better? Drastically?

Here's an article on AlphaGo and the machine that taught itself to win every time.

Five months later, we received another Nature paper for AlphaGo Zero. Unlike the earlier versions of AlphaGo which learnt how to play the game using thousands of human amateur and professional games, AlphaGo Zero learnt to play the game of Go simply by playing games against itself, starting from completely random play.

In doing so, it surpassed the performance of all previous versions, including those which beat the World Go Champions Lee Sedol and Ke Jie, becoming arguably the strongest Go player of all time.We believe this new breakthrough has the potential to facilitate major scientific breakthrough and in doing so drastically change the world for the better.

https://deepmind.com/research/alphago/

[jbuck919]

Go was the last resort for elite mathematicians after it was discovered that no human could beat the best chess program. Now even go is gone. Now mind you, neither of these games has been "solved" in a mathematical sense, as was checkers long ago. (It is impossible to win at checkers against a properly programmed computer.) It is just that the computers play them better.

I would recommend the book Rise of the Robots by the ironically named Martin Ford.

It was Bertrand Russell who foresaw many years ago that the day would come (he thought soon) that people would only have to work about four hours a day and the rest would be comfortable leisure. One of the main reasons this has not already happened is because of the tyranny of personnel management. Nobody wants to pay someone decently for only four hours of work a day. Myself? I could never fill an eight-hour day and pretended for years to be busy when I could get done in two hours what it appeared to take others a whole day to do. (That kind of work, incidentally, has been entirely taken over by computers, a process to which I actually once contributed.) That's why I changed careers to become a teacher, but I traded unutterable boredom for a different set of problems. Now I'm happy to be semi-retired, and just hope Trump & Co. don't take that away from me.

[Belle]

These kinds of U-tube clips (following) really demonstrate the extent to which the nature for work has changed by technology and robots in the auto industry. There are people still needed to work, but this involves ever higher skills. We've just had the car industry completely close in this country - within the last week - and a great many people in that and subsidiary industries have now lost their jobs. This is a live issue around the world. Dr. Jordan Peterson says that for people in the lowest IQ percentile the main source of work is driving, and what is on the horizon but 'driver-less' vehicles.

After we recently bought our new (German) car we became interested in the modern production line and we can see for ourselves the result; panels which perfectly align and a reduced number of sections in the body of the car which facilitate robot construction. Gone are the days of rattles! Everything is as tight as a drum. When we picked it up and the salesman showed us the engine I asked where the battery was and he had to yell out, "hey Bob, where's the battery?" and it was hidden under the driver's seat - apparently it doesn't need attention.

While some of these robotic machines strictly probably don't conform to the idea of 'robots' having artificial intelligence, they are still capable of completing quite complex actions. This is very interesting:

https://www.youtube.com/watch?v=cNxaXUdE38U

As is this, from about 2:30.

https://www.youtube.com/watch?v=GlZ8BimtqOA

There's more human engagement with this one:

https://www.youtube.com/watch?v=PMB5a4Btve0

Compared to:

https://www.youtube.com/watch?v=JkCJkdcl8mw