An AI Lab Will Fix Your Plumbing for Free — to Record It for Robot Training

October 19 — An AI lab focused on collecting real-world data plans to work with homeowners to record skilled tradespeople at work, in order to train AI robots. The company, called HUMXN, says it will subsidise plumbing, electrical, HVAC, duct-cleaning and carpet-cleaning services: residents who need such work can get it free, in exchange for allowing people wearing “wearable capture devices” into their homes to record.

HUMXN founder Thomas Unise said: “What the next generation of AI needs will no longer be just text and images from the internet. Robots need to understand how professionals interact with the physical world. Home-service technicians complete millions of highly skilled jobs every day. We built this model so that homeowners and service businesses can earn economically from the value of that expertise.”

He added: “Millions of skilled workers carry out extremely complex tasks every day, but very little of that expertise exists in a standardised data form that AI systems can learn from directly. We want to build the infrastructure that connects this kind of knowledge with companies developing the next generation of intelligent devices, while making sure the people who create the value share in the returns.”

HUMXN says it is hard to capture technicians’ natural behaviour in a laboratory. By working with homeowners, it can collect data from different houses, equipment types, fault scenarios, individual technicians and environmental conditions. That data lets AI models learn the entire process of a technician handling a job: surveying the site, testing the system, choosing the right tools, dealing with surprises and judging whether the work is acceptable.

The lab says the scheme lets homeowners get qualifying home services for free while ensuring service workers are paid as usual. In essence, HUMXN pays residents in exchange for permission to record video and collect real training data inside their homes. The project starts in three cities — Minneapolis, Chicago and Miami — to capture data from different climates. If you live in one of them and need repairs or services at home, you can visit HUMXN’s free home service page to see whether you qualify.

Meta, Nvidia, OpenAI and other AI labs have all been accused of scraping data to train models without permission; by comparison, HUMXN’s model of paying directly for training data is quite novel. But the next generation of AI probably cannot be trained on scraped text, images and video alone, so companies like this may have to pay for data generated in the real world.