It's a long read but interesting if you're up for it. That being said I want to comment and disagree with two things that were said in the last section titled "Coda".
1) "AI does teach us about intelligence. It teaches us that “intelligence” is a motley assortment of heuristics, kludges, and cheap tricks." I wanted to point out that he seems to be making a fatal assumption; because we have solved problems that were considered to be in the domain of AI and the solutions to those problems were "a motley assortment of heuristics, kludges, and cheap tricks," a true general AI (Strong AI) will look the same, and therefore that's all our intelligence is. The author pokes a bit of fun at other people who have made grand claims and predictions about AI but then he himself goes and does the same thing. The truth is we really don't have a good idea about what a real general artificial intelligence will look like and based on the lack of progress I'd guess the chances of deriving a solution from known methods aren't good.
2) "If we were trying to build a true, general AI, we would first need to create a way for it to get around and interact with the larger world." This is another instance of the irritatingly pervasive idea that we need to have an AI acting in the real world with a real "body". I can not understand why some people see this as so crucial. To me it seems to do nothing but add a whole lot of unnecessary complications to an already very hard problem.
Assumptions like these are never going to get us to a general AI. People need to stick to observable facts, make theories based on them, and then test those theories.
No comments:
Post a Comment