How computers beat humans at chess
My son turned four yesterday.
A couple of weeks ago, he was with his older cousin, who became really good at chess, and wanted to play against everyone. So my son watched this new game being played in front of him, and he started to be intrigued. More than anything, the threat of saying « Check » seemed to be an exciting perspective.
He has been talking about chess since then, so yesterday, his birthday gift was a small chess board. On the box, it said: « Object of the game: Capture enough of your opponent’s chessmen to take your opponent’s king into a checkmate position. »
(…)
AlphaZero is the most advanced computer program to play chess. Instead of being taught chess through exposure of the thousands of games played by the best human chess masters, AlphaZero discovered chess from scratch, playing games against itself.
Today, AlphaZero can beat any human being at chess. To understand where his strategic strength comes from, an analysis of AlphaZero’s gameplay has been made. One major insight came up: it seems like, after doing all its analyses, AlphaZero is not obsessed about capturing opponent’s chessmen as much as the best human chess masters. The common wisdom about chess was that, the more chessmen you capture from your opponent, the easier it will be for you to corner his king into defeat. But, because AlphaZero didn’t learn chess from humans, it didn’t pick up this piece of wisdom. It just focused on the most effective way to win a game, as proven by millions of simulations. And, it turned out, capturing your opponent’s chessmen does not rank as high as human thought.
(…)
The fact that my son’s chess box used that incorrect piece of wisdom as a way to describe the purpose of chess is fascinating. It’s the perfect metaphor for so many situations: Offering advice to someone, framing a problem to be solved without hinting at a particular solution, wondering about the long-term damages of short-term thinking, giving a fish vs teaching out to fish vs showing the river, etc.