Very
recently, South Korean Lee Se-dol (the world's greatest “Go”
player) was beaten 3 out of 5 games by the AI, “AlphaGo.” The
game of Go has been played by humans for over 3,000 years, and AI
mastery of the game was believed to be at least a decade away. Yet
AlphaGo taught itself to play the game by playing it against itself
millions of times to learn and correct its weaknesses. The AI was
able to teach itself from an amateur player to a world champion in
less than a year. Some have projected that the singularity—the
event when AI surpasses all human control and understanding—is
within our lifetime, with dates ranging as early as 2027 to as late
as 2045. Considering previous projections about AI (AlphaGo, for
instance, projected to take ten years longer than it did to achieve
Go mastery), it is reasonable to assume that the current rate of
progression in AI technology is going to accelerate beyond our
expectations. We need to do something, now, and not be reactive to a
situation we may find has quickly moved beyond our control. I've
always considered science fiction as a warning. Are we paying
attention?
No comments:
Post a Comment