HUMXN offers free home repairs in three cities to gather AI training data

AI Lab Seeks Real-World Data

HUMXN, an AI lab, is partnering with homeowners to record skilled laborers performing various home services. The goal is to gather real-world data to train AI robots, as current AI systems lack understanding of how skilled individuals interact with the physical world.

The company will subsidize services such as plumbing, electrical, HVAC, air-duct cleaning, and carpet cleaning. Homeowners in select cities can receive these services for free in exchange for allowing technicians equipped with "wearable capture technology" to work in their homes.

Bridging the Gap in AI Training

Thomas Unise, founder of HUMXN, stated that the next generation of AI requires more than just text and images. He emphasized the need for robots to comprehend how skilled people perform technical tasks in the physical environment. Millions of complex tasks are performed daily by home-service technicians, but this expertise is not readily available in a format AI systems can learn from.

HUMXN aims to build infrastructure that connects this knowledge with companies developing intelligent machines, ensuring that those who create the value also benefit economically.

Capturing Diverse Scenarios

The lab noted the difficulty of capturing natural movements of skilled workers in controlled lab settings. By collaborating with homeowners, HUMXN can collect data across diverse environments, including different homes, equipment types, problems, and technicians. This approach provides AI models with insights into the entire process of a technician's work, from assessment to tool selection and task completion.

Program Launch and Scope

HUMXN's system ensures that homeowners receive qualifying work for free while service professionals are still compensated. The company essentially pays residents for the right to record on their premises. The program is launching in Minneapolis, Chicago, and Miami to gather data from areas with varied climates, which will provide a broader range of scenarios for AI training.