Sony AI's Ace robot beats elite players in table tennis

12 articles · Updated · New Atlas · May 4

Detailed in a Nature paper, Ace rapidly learned and made split-second adjustments, including returning net-clipped balls in real time with aggressive, accurate shots.
Researchers said it is the first real-world autonomous system competitive with elite human table tennis players, a sport chosen for its demands on speed, precision and adversarial play.
The demonstration highlights advances in self-correcting robotics and suggests AI systems are moving from surpassing humans in games to challenging them in fast physical sports.

After mastering table tennis, which complex human sport will AI conquer next?

Can a machine that only learns in simulation ever truly outwit a creative human athlete?

How will this self-correcting AI technology reshape jobs that require split-second physical decisions?