Google AI Enjoys “Amateur” Table Tennis

Google DeepMind has developed a groundbreaking table tennis-playing robot that could serve as a formidable opponent, particularly for intermediate-level players.

The London-based team behind the artificial intelligence (AI) project aimed to create a robot that could provide a competitive and enjoyable experience for human players. The result might not be Olympic-ready, but it marks a significant advancement in robotics and AI integration.

Using a combination of a skill library and a high-level controller, the robot selects the most effective techniques in real time, adapting to the game’s progress and the opponent’s skill level.

The AI was initially trained using a small amount of human data, and further honed its abilities through simulations and reinforcement learning. It then faced off against 29 human opponents of varying skill levels, winning 100% of its matches against beginners and 55% against intermediate players, though it lost to advanced players.

Despite its limitations, the robot was praised for being “fun” and “engaging” by all opponents, even those who exploited its weaknesses.

As a training tool, it shows great promise, potentially outperforming traditional ball-throwers in coaching scenarios. This development highlights the ongoing advancements in AI and robotics, offering a glimpse into the future of human-machine interaction in sports.