Tuesday, October 24, 2017

AI Unhindered by Paradigms (part 2)


Several years ago Google's AI, known as DeepMind, learned to play the strategic game "Go" and beat the world champion Go player. The AI program called "AlphaGo" has since been playing against itself, developing moves and strategies that expert Go players are still trying to understand. The moves this AI has been making have been described as totally alien, as if knowledge from the far future has been dropped on earth, much of which is still being analyzed as experts are attempting to comprehend it. The AI is doing things that Go players have never done, even doing things that they thought were bad moves or mistakes yet turned out to be brilliant in the end. Sometimes even its opening moves are so confounding that players and analysts have given up trying to understand it--it's so bizarre and seemingly senseless. Yet the AI continues to win...

No comments:

Post a Comment