It’s funny how some things always seem to be in the future.
This is perhaps more true with artificial intelligence (AI) than anything else.
Despite the fact we use and/or see AI in action pretty much daily, it still seems like it is far away.
Part of that is from moving the goalposts. It was 1997, almost 20 years ago now, that IBM’s “Deep Blue” beat world chess champion Garry Kasparov at his own game.
At the time, it seemed a major breakthrough, but ultimately, for most experts, Deep Blue’s accomplishment did not demonstrate intelligence.
Deep Blue basically used brute force to best Kasparov. Capable of calculating more than 100 million chess positions per second, the supercomputer merely overwhelmed the champ with its superior computing power.
Similarly, when IBM’s “Watson” beat two of the very best Jeopardy champions in 2011, it was a touted as a big win for AI.
Certainly, Watson was a big step up from Deep Blue in that it is capable of answering, in near real-time, questions (or providing questions for answers in the case of Jeopardy) posed in natural language.
They call Watson a cognitive learning system. It did, as a matter of fact, learn the game by practicing.
Critics, however, view it as an impressive trick based on number-crunching, pattern-matching and database-searching and does not demonstrate true thought.
According to Patrick Winston, former director of MIT’s AI Laboratory, what Watson cannot do is connect life experiences to form cohesive thoughts.
He says, for example, that Watson would not be able to attend a conference about Watson because it would not be able to participate in discussions about itself and how it works.
Self-reflection and self-awareness are something that we are very reluctant to ascribe to anything besides ourselves.
As machines get smarter, we tend to tinkle with the definition of intelligence.
Semantics, of course, are one of the big things that trip up computers. Until they get that. Until a computer actually becomes indistinguishable from a human being, what we call it will probably always be controversial.
With each little step, however, we get closer to that possibility.
