Reading 11: Is AI cool or spooky?

I would consider AI any application of computer intelligence that attempts to imitate human intelligence and can become smarter over time. I’m simultaneously very impressed and slightly unnerved by the descriptions given in the readings of the AI programs that have been able to beat humans in various games. I always have kinda shrugged at the claims that AI will take over the world, and I still do a bit, but I was a little spooked when the articles described times when even the people who created these programs couldn’t explain their decision making. I’ve heard in other classes about unsupervised learning, when programs are able to categorize information without labeled training data, and I always thought that it was just a nifty way to avoid labeling. However in these articles, I began to see unsupervised learning as a spooky way for computers to learn things that we haven't even taught them — and possibly the first step toward something we no longer have control of.

I’ve always thought that the Turing test is a little silly. I just think that there should be many more factors into deciding if something is truly “intelligent” than being able to fool people into thinking it is human. It seems like an oversimplified and unscientific answer to a really important and complex question. I also think though that the Chinese Room argument is a poor counterargument because what we mean by saying a computer understands is not the same as when we say a humans understands. A computer will never have thoughts in the way that humans do, but can only give comprehensible output given input.

Despite being definitely spooked by some possible implications of AI, I’m really not worried about any of that becoming reality — at least not anytime soon. I think that exploring various areas that AI can help us make more informed decisions or automate processes is beneficial to society. To me, nothing is scary about AI that gets very good at accomplishing a single defined task and isn’t designed to understand anything beyond that scope. If we started to combine these applications I might get a little concerned but I don’t see that happening soon.