At the first International Humanoid Olympiad in ancient Olympia, Greece, humanoid robots demonstrated basic skills like playing soccer, shadow boxing, and shooting arrows—movements that were often clumsy and interrupted by battery changes. Despite their limited capabilities, the event highlighted a growing ambition to bring human-like robots into everyday life. Organized by Greek academic Minas Liarokapis and others, the four-day gathering brought together developers, researchers, and futurists at the historic site where the Olympic flame is lit. The central question echoed through the halls: when will robots be capable of performing simple household tasks like tidying rooms or doing laundry? Experts agree that physical robots are still far behind their AI counterparts. While artificial intelligence has advanced rapidly thanks to vast digital datasets, humanoid robots face a major bottleneck—there is little real-world training data available. Unlike text or images, physical actions are slow, expensive, and difficult to record at scale. According to a recent article in Science Robotics, humanoids are roughly 100,000 years behind AI in terms of learning efficiency from data. Ken Goldberg, a professor at UC Berkeley, argues that progress will require moving beyond simulations and embracing real-world experience. He suggests robots should collect data while performing practical tasks—like delivering packages or driving taxis—to accelerate learning. Collaboration is key. Luis Sentis, a professor at UT Austin and co-founder of Apptronik, noted that major investments are flowing into robotics, driven by partnerships between researchers, data firms, and manufacturers. These alliances are beginning to solve core challenges in mobility, dexterity, and autonomy. Some innovators are finding creative ways to bridge the gap. Aadeel Akhtar, CEO of Psyonic, which builds bionic hands with sensory feedback, shared that data from human users is being used to train robots. “We’ve built our hand for both humans and robots,” he said. “That’s how we’re closing the gap.” Meanwhile, Hon Weng Chong of Cortical Labs is exploring a radical approach—using living brain cells grown on chips to create biological computers. These neurons can learn and adapt, potentially giving robots a more human-like way of processing information. The Olympiad focused on achievable tasks, avoiding events like high jump or javelin throw, which would require specialized hardware. Still, organizers aimed to create a platform for honest, annual evaluation of real progress. China has emerged as a leader in public robot showcases, with events like Beijing’s first Humanoid Robot Games in August. In contrast, U.S. companies often rely on polished videos that hide technical shortcomings. While some U.S. teams attended the Greek event, few brought physical robots. Elon Musk’s Tesla Optimus made a splash in 2022 with a stiff, robotic walk and wave—though its performance was far from perfect. Boston Dynamics has also drawn attention with its Spot robots dancing to Queen’s “Another One Bites the Dust” on America’s Got Talent. When one robot broke down mid-performance, judge Simon Cowell remarked it was oddly beneficial—proof of how difficult the technology remains. The journey from ancient Olympia to the modern home is long, but the race is on.