Meet Ace, the table-tennis robot that can beat elite players

This episode of The Nature Podcast explores two groundbreaking scientific developments: the creation of ACE, a table-tennis-playing robot developed by Sony AI, and the latest challenges in measuring the gravitational constant, Big G. ACE, designed by Peter Dörr and his team in Zurich, combines advanced perception systems—using high-speed cameras and event-based sensors—to track the position and spin of a ball moving at up to 30 meters per second. Trained entirely in simulation, the robot’s neural network learns to play with precision and adaptability, culminating in victories over elite human players, including a top-25-ranked woman and a top-200 man. The robot’s success highlights breakthroughs in real-time sensing, adaptive control, and custom hardware. Meanwhile, the episode delves into the enduring mystery of Big G, the gravitational constant, which remains poorly understood despite centuries of effort. A decade-long replication of a French experiment by NIST scientists—conducted with full blinding to avoid bias—produced a value significantly different from both the original and the internationally accepted average, underscoring the extreme difficulty of measuring gravity with high precision. Experts like Lizzie Gibney and Esther Colombini reflect on the broader implications: while Big G may not currently impact practical applications, the pursuit of its value drives innovation in metrology and inspires fundamental scientific curiosity. The episode closes with a call to appreciate the value of challenging, seemingly niche research that pushes the boundaries of human knowledge.