When Robots Attack

How can an AI can play masterful chess, yet be so clumsy it breaks a kid’s finger? “Moravec’s Paradox”

If you follow nerd media you may have seen this clip above — of a chess robot accidentally breaking the finger of its opponent, a seven-year-old child.

It’s quite distressing to watch! The robot is moving its pieces around, and the child reaches to grab a chess piece — apparently before the robot had finished its move. While continuing its move, the robot grabbed the boy’s finger, strongly enough to seriously hurt it, and to force a cluster of panicked adults to rush up and free the child.

I saw comments on Twitter with people wondering, essentially, how in hell can a robot play chess so well yet be incompetent enough to not realize it’s breaking a kids’ finger?

As it happens, there’s a useful concept in robotics that helps explain this — and that also highlights the really big challenges facing today’s AI.

Step forward, “Moravec’s Paradox”.