This is no small thing: Google’s “Alpha Zero” AI, after taking just 4 hours to teach itself chess, played 100 games against the strongest dedicated chess engine, Stockfish, with decisive results: it won twenty-eight games, drew seventy-two, and lost… zero.
We had a good run, humans.
One Comment
But don’t compare this to a human. This form of AI is still essentially hierarchical and linear. It has parallel streams of calculation going, but still is simply searching all the possible paths. Having studied both wet and electronic brains, all the AI fears are just that, fears. The only ways machines will replace humans is in jobs that can be expressed as an algorithm. That means many of the repetitive-task jobs will be gone. The challenge in AI is to find adequate work for those of limited learning ability.