Researchers at the University of Tokyo have developed a robot that cheats at rock-paper-scissors
by detecting the gesture you’re about to throw. It’s the automated
equivalent of your jerk friend hesitating a moment before committing to
their move — except that it happens at superhuman speed.
R-P-S is a great casual decision-maker. But the game only works if
you and your opponent reveal your choices at the same time. Here, the
robot perceives and acts faster than the human eye; it’s onto you as
soon as your hand begins to form a shape. By the time you’ve landed your
move, it’s already countered with the winning one.
It’s a neat AI tech demo, but it also holds an important parable for
how to think about intelligence. Rock-paper-scissors is a trivial game,
and yet we’ve only decided to focus our energy on it a decade after cracking chess.
Chess used to be the brass ring of AI research, but projects like the
R-P-S robot point to a different kind of intelligence — a physical
kind. We’re used to thinking about intelligence in terms of brains
versus brawn, but as it turns out, making machines that can exercise
their muscles takes a lot of smarts, too.
As more digital things cross over into the physical world, it’s that
physical intelligence — the ability to move and perceive in space —
that’s going to really make the difference.
0 σχόλια:
Post a Comment