The request was deceptively simple: film yourself performing mundane domestic tasks—placing food in a bowl, microwaving it, removing it—in exchange for cryptocurrency. Elsewhere, users are being recruited to play games that involve remotely controlling a robotic arm in Shenzhen, China, to solve puzzles. These are not merely oddities of the gig economy; they represent a desperate search for the "physical intelligence" required to make humanoid robots a reality.
Just as the written word fueled the rise of large language models, the nuances of human movement have become the latest frontier for data collection. While AI companies built ChatGPT by scraping the vast, pre-existing archives of the internet, roboticists face a much steeper climb. There is no "internet of movement" to download. Until recently, companies relied on virtual simulations to teach robots how to navigate the world, but these digital environments often fail to account for the messy physics of reality—the subtle friction of a surface or the specific elasticity of an object.
This "reality gap" has led to robots that literally stumble when they leave the lab. To bridge it, robotics firms are now betting on scaling laws: the idea that if you feed a model enough data, it will eventually find the patterns of success. By harvesting the way humans move through their daily lives, these companies hope to create humanoids that can seamlessly slot into our existing infrastructure, performing the labor we no longer wish to do.
With reporting from MIT Technology Review.
Source · MIT Technology Review
