Since robots began outperforming humans at sports, public concern about artificial intelligence has grown significantly, with Americans expressing skepticism about AI's benefits and calling for stricter government oversight. Multiple policy bodies, including the White House and state governors, have begun developing regulatory frameworks to govern AI development, while experts in countries like New Zealand have criticized government approaches to managing the technology. These developments reflect a broader shift from celebrating AI achievements to focusing on how to safely control and regulate these increasingly powerful systems.
Since robots mastered tennis and ping pong, the AI robotics field has expanded into new domains: medical screening now uses AI to detect blindness in babies, while companies are developing AI chefs and insuring physical robots against damage. Additionally, major tech investors are backing AI robotics startups, with a $13 Nvidia-backed AI stock recently partnering with Anduril, a defense technology company, signaling growing commercial and industrial applications beyond sports.
Sony and AI researchers have built robots that teach themselves to play sports—and they're now better at tennis and ping pong than most humans.
Here's what makes this different from previous robot tricks. These robots don't memorize movements from humans or follow pre-programmed instructions. Instead, they watch what happens after each paddle swing, learn from mistakes instantly, and adjust their next move on their own. Think of it like a child learning to ride a bike without being told the exact angle to lean—the robot figures out the physics through trial and error.
Why should you care? This skill—called reinforcement learning (training yourself by trying things and seeing what works)—is the same method robots will use to learn jobs that don't have step-by-step instructions. Factories, hospitals, and warehouses rely on tasks that change constantly. A robot that can adapt in real time is worth far more than one that can only do one job the same way forever.
Researchers at NVIDIA and Physical Intelligence have also shown that robots can now watch other robots work and copy new skills they've never been taught. One robot sees another doing something new, understands it, and then does it itself. This spreads knowledge between machines like humans sharing tips with coworkers.
The real breakthrough? These robots are becoming reliable at handling unexpected situations. A tennis robot doesn't just hit balls coming at it the exact same way every time. It adjusts for spin, speed, and angle—skills that require actual understanding, not just memorized moves.
For workers: this matters because jobs involving repetitive tasks will change faster. For businesses: robots that learn independently could cut training time from months to weeks. For society: we're moving toward machines that don't need constant human supervision.
What you should think about: If your job involves learning new tasks often or solving problems that change, your skills in adapting quickly become more valuable than ever—that's something robots still can't do better than humans.