Micro AGI is dispatching free cleaning crews and a private chef into New York City apartments to record workers' hands and object use. The data will train household robots. Separately, prediction market Polymarket now assigns a 95% probability to Anthropic's AI capabilities.
How the Data Collection Works
The company is offering these services at no cost to residents. Cameras capture every hand movement, from gripping a sponge to chopping vegetables. The data includes which objects are picked up, how they are held, and the sequence of actions. Workers go about their normal routines, but their hand movements are tracked in detail. Micro AGI has not disclosed how many apartments are involved or how long the program will run.
Why Hands and Objects Matter for Robotics
Teaching robots to manipulate everyday objects requires detailed observation of human hand movements. By recording workers in real home environments, Micro AGI hopes to gather naturalistic data that simulations or labs may not provide. The sequence of actions, grip types, and object interactions are all recorded. Robots today often struggle with tasks that require fine motor skills. By observing human hands in natural settings, the company aims to train models that can handle a variety of objects and environments.
Market Confidence in Anthropic
On Polymarket, the probability that Anthropic's AI capabilities reach a certain level has climbed to 95%. The platform's users are betting heavily on the company's technology, though the specific capability in question is not detailed in the market's description. Polymarket is a decentralized prediction market where participants trade contracts based on event outcomes. A 95% probability is unusually high, indicating strong consensus among bettors.
The recording initiative is underway in select NYC apartments. Micro AGI hasn't said when the collected data will be used in a commercial product. The market's odds on Anthropic reflect a broader interest in AI progress, but the two developments remain separate.
