Quick note if you missed this: the Google DeepMind team now has AlphaGo Zero. This new AI learned Go from scratch, and is badass. They had it play against the same version of AlphaGo that beat Lee Sedol last year... and it won... 100 games to 0.
Just to recap: that older version of AlphaGo beat a world grand champ human 4 games to 1. Basically we're all just John Henrys now, but don't even get that moment of victory before kicking the bucket. It took it 3 days of trial-and-error game learning to surpass the Lee Sedol level.
Of course we already decided that humans facing off against humans is more fun. Sure, an army of Boston Dynamics monsters could probably school the Pats all over the field, but that doesn't mean human football has lost any of its clout. And some higher end animatronics can out ballet us. There's even a golf club swinging arm-on-a-cart out there with radar laser vision something or other calculating the right trajectory and force for each swing and nailing holes-in-ones. We already ceded our physical skill to the machines. It makes sense they'll beat us in strategy games. But games are games. They're fun. And humans fighting robots is really only fun at first for novelty, and then later in movies when they're trying to terminate us. Now that will be the match to keep an eye on. Meanwhile we're still waiting on the supposed breakthroughs in medicine and science that system runners like AlphaGo's keep promising. That's my ten won, anyway.
Check out more on AlphaGo Zero here:
Comments