Chess Thoughts. Hugo Thoughts, Too!

I’m working a bit under the weather this week, so: here’s a pretty great video which you may have seen elsewhere.  Chess grandmaster Maurice Ashley challenges a NYC chess hustler to a game, without revealing his true identity.  Got to wonder what the hustler thought about being on camera, but setting that aside, it’s a great clip and well worth your five minutes.

I first saw this video on Boingboing a while back, but when Shut Up and Sit Down reblogged it, they added this link to the actual game played, which is a whole different kind of interesting.  My last exposure to computerized chess was, god, a little over a decade now.  As you walk through the game (right and left arrows move you forward and back, respectively), note how the move list in the left sidebar indicates when each player makes a mistake—not a rules mistake, to be clear, but a tactical or strategic mistake, according to the computer’s calculation.

Also really cool: the health bar beside the board, and the graph beneath, registering the positional advantage of black and white.  You can actually see, move by move, how white loses!  It’s one thing to know, in theory, that positional chess play requires developing pieces and controlling the center of the board.  It’s another to see white take a huge dive in the graph at move 15 when they play Nh2.  This sort of thing really makes clear why people are excited about the Alpha Go result—better computer play offers human players a deeper understanding of a beloved game, and develops the art overall.

(Food for thought, though no guarantees about nutritional content of said food: to what extent is a computer capable of placing the correct moves in a Go game, or a chess game, actually performing the activity humans reflexively describe as “playing go”?  A professional chess player develops patience, mental endurance, and profound mental habits required to bend her omnivore-scavenger brain to the profoundly non-omnivore-scavenger activity of staring at a game board for several hours at a time, oblivious to any potential predators creeping up behind.  These are additional “rules” to the game as played by humans—or at least, they’re constraints to which human players are subjected.  “Learning to play chess,” for a human, is really “learning how to navigate human embodied cognition in such a way as to win a chess game.”  Is a hydraulic car-moving robot stronger than a champion weightlifter?  On paper it can move more weight.  But I suspect we use the word “strong” to mean different things in different contexts.)

(In case this isn’t clear, what I’m not doing here is attempting to qualify away AlphaGo, or computational chess playing, or hydraulic car-moving robots.  They’re all obviously accomplishing the tasks for which they were designed!  There’s no room in a checkmate for qualia.  But along the way, I think developments in artificial intelligence reveal unexamined assumptions about the nature of the tasks they’re designed to confront—they force us to ponder the context of thought.)

(I suppose I for one am supposed to welcome our new robot overlords at this stage in the conversation, aren’t I?)

Setting that aside, news!

Thanks as ever.  Also: check out today’s episode of #ColdWitch!

3 Responses to “Chess Thoughts. Hugo Thoughts, Too!”

  1. Laurence Brothers

    > “Learning to play chess,” for a human, is really “learning how
    > to navigate human embodied cognition in such a way as to win
    > a chess game.”

    This is a key insight for human-chauvinists to cherish for the time being. Of course a diagonal cutter is better at snipping wire fences than a human finger, but it’s humans who build diagonal cutters for their own use, just as it’s humans who build and program computers. When a naive computer program with no inherent notion of games and lookahead and so on learns to play chess or go, then we can worrying about stepping aside for them….

    And wouldn’t it be sad if the first fully self-aware AI systems have to spend so much of their processing power solely on consciousness and on real-world knowledge management that they aren’t very smart at all? Perhaps the first demonstrably conscious AIs will only be doglike or crowlike in their cognitive skills, or they may struggle to perform even routine human tasks. It will be a rather pathetic existence for those early AI generations, if this indeed turns out to be the case.

  2. Kellan Sparver

    One of my favorite complications to these game AIs is to suggest that, okay, it’s great that they can beat human players while pulling down tens or hundreds of kilowatts of power—but can they do it on 20 watts (the rough power draw of the human brain)?

  3. AntVicino

    It’s interesting the lichess engine doesn’t outright classify Nh2 as a blunder. White concedes not only the center, but also the tempo, by removing his piece from the field of play for, at minimum, two moves. Perhaps not a game changer in recreational play, but against a Grandmaster there is simply no recovering.


Leave a Reply