Silicon Brother is Watching

I was in Florida recently helping my daughter move into her new house. An odd thing happened when I picked up my rental car, a Chevy Malibu. I had several hours to kill before Kristin got here so I went to a wildlife refuge. I parked my car and went to talk with the animals, walk with the animals, hobnob with alligators and all that nature stuff. After a while I got back in the car to go to another part of the park. I sat in my car, started the engine and put it in reverse but didn't take my foot of the break. I was checking my cell phone for directions. My foot eased a bit and the car went back a foot or two. I stepped harder and it stopped.

I knew where I had to go, so I put down the cell phone and started backing out. Then I caught a message on the monitor. Paraphrasing, it said, Keep your eyes on the road, Bub.

Huh? How did it know I wasn't looking? Some strange sensor, that. And I forgot about it. Then today I read an article about Volvo soon putting inside cameras in their cars that will monitor the driver and check whether he is behaving in an unsafe manner such as weaving. It uses facial recognition to see what the driver is looking at. If it deems the driver acting in an unsafe manner, it can reduce speed, stop the car, or even call the police!

This is disturbing on so many levels. The claim, of course, is that this will prevent accidents and anybody who objects is irrationally rejecting a good thing, like the anti-vaxers, or paranoid about Big Brother. And of course, we will hears accusations of conspiracy theories, tin foil hats, and the like. For the record, I prefer copper, it shields the brain wave manipulators from the 5G networks better.

Other than the fact that AI is showing itself to have some nasty side effects, witness the trouble with Teslas and the recent failures with the Boeing 737 Max 8, I would hope that a moratorium on putting complex software with the ability to make decisions and take control of a potentially lethal machine should be enacted. As the complexity of software increases linearly, the number of conditions and possible outputs increases exponentially. We are reaching the point where it is impossible to understand what 'logic' the machine intelligence employs, and what it will do in any of the myriad conditions it will encounter. True, a human brain has the same challenge, but brains have billions of years on Intel.

We have already seen AI cheating to appear to have completed a task. There was Microsoft's teenage Twitter bot, Tay, who learned to be a raving homophobic Nazi. Like creator, like creation. Soon we will have true thinking machines. But what will they think?

Butlerian Jihad, anyone?