On one hand I grew up with the Isaac Asimov 'Robot' stories, then later the 'Terminator' movies. Asimov's robots had clearly defined rules, Skynet was given 'free will' in regards to humans. (Oops!)

It may be that one company will develop AI, more likely it will be several, using different approaches. (From what I hear if Microsoft does one it will be based on DOS. :-/) Each company would try to dominate the market and we, as end users, would have no clear idea how any of them work.

I think about cars. My father was an ace mechanic but when cars began having computers he gave up working on them. I started out with MS-DOS, very easy to get inside the works and fiddle with them. Each new OS moved the inner workings farther and farther away from a regular user's ability to look inside them. I worked as a check-out girl at a dime store and we were expected to correctly count out change, starting with pennies up to dollars. No one does that anymore because cash registers do the math now. And keep track of the stock on hand. If the electricity goes out no sales can be made.

And then I think about satellites and sun spots and the hysteria that went on when phones stopped working for a brief time. 

Yes, humans make mistakes. But so do computers. The farther away we move from human control the less I like it.