>>512816743"AI" is good at one thing and one thing only. That is, showing you stuff that's really hard for humans to learn.
Most people play chess or go knowing the game as their grandfather explained it to them, their grandfather being a casual player and explaining it to them poorly.
Machine learning BTFO's old game playing models because old game-playing models relied on playing like retarded autists who "keep score" and only take mostly arbitrarily defined "optimal plays" with perfect computer discipline.
The "Ai" bot on the other hand understands the game from a pure mathematical standpoint, the game is one of area denial, and denying area depends on a fixed set of variables, some of which are too subtle for anything less than a master of the game to even comprehend or perceive, perhaps some beyond even them. "Ai" does perceive them, and given billions of dollars of training (more than any human's life is worth) it can memorize and quantify all of them. It runs this calculation that cost literally hundreds of lives, to produce the modest result that humans think is paradigm-shifting. That is, "white pawn to F6 is a pretty good move."
It's cool yea... but it takes near perfect play and stone cold computer nerves to get to that point. Humans choke trying to do it. Which is why grandmasters never learned it, and in hindsight, it makes lots of sense. White pawn in F6 is a nasty little troll, it denies area, especially if the rook is still there which is something you should have already known about chess if you cared enough to play beyond childhood.
So, yea, 10/10 pretty cool edgecase, as a species we are now a billion dollars poorer and the average ELO of the human race is maybe 25 points higher because Ai explained and demonstrated chess better than old peepaw ever could.
I suppose knowledge is priceless after all