Rise in humanoid robots creates need for unusual new job

The dream of deploying humanoid robots in every home has created a new type of job,  in which people record themselves doing mundane household tasks.

With the evolution of AI, robots resembling humans have become the latest frontier in the march ahead of new technologies, but as a CNN article points out, this requires a vast amount of data to learn how to safely and effectively replace humans.

This constant need to be fed sets of data has created an appetite for footage recorded in the first-person that can be used to train robots, which is known as “egocentric data”, or “human data”.

According to the outlet, startups have stepped up in several months to supply that demand by collecting those videos from thousands of contract workers around the world.

“Manufacturing, factory warehouses, retail, nursing homes, hospitals – you’re going to need this type of data in basically every single environment, and that’s because the movements are all different,” said Arian Sadeghi, Vice President of robotics data at Micro1, which began recruiting its own team of remote videographers last year.

They equip each person with a headgear to attach a camera and provide them with filming instructions and a list of tasks such as cooking, cleaning, gardening and pet care. Workers are expected to submit at least 10 hours each week, in which they alternate between assignments.

For now the firm is focused on household chores, but Sadeghi said they encourage contractors to experiment with what they film “in case it would eventually help robots adapt more quickly.”

Though Micro1 is based in California, it employs about 4,000 “robotic generalists” across 71 countries. This allows the company to collect approx. 160,000 hours of video each month, but Sadeghi says that this is nowhere near enough.

“You need probably billions of hours,” he said. “We haven’t even gotten to human interactions. This is just simple household chores.”

According to the company owner, the growing demand for data in robotics mirrors the early trajectory of AI chatbots like ChatGPT, which were trained on hundreds of billions of words harvested from the world wide web.

Those AI models evolved by relying on readily available content online, but robot development requires a much more specific set of training data which has no instant library to reach out to.

By Nazrin Sadigova