Shift will clean your house for free. The cost: data captured by a camera worn by the cleaner, which will be used to train humanoid robots.
John Koetsier
Shift will clean your house for free. The cost: data captured by a camera worn by the cleaner, which will be used to train humanoid robots. “Tomorrow’s robots learn from today’s work,” says the startup, which employs robot trainers worldwide but is launching its home-cleaning service in New York.
People are taking advantage of the opportunity.
“A vetted shift operator comes to your home wearing one of our devices. They clean. They leave. You pay nothing,” the company said in a posting on X yesterday. “In return, we record the cleaning. The robotics is based on data about how people do everyday tasks, and the value of that recording is what funds the service.”
The post already has thousands of likes and retweets and more than 1,000 replies, thanks to Shift asking users to comment in exchange for receiving an early access link to the service.
The challenge, of course, is privacy. You might be wondering: is my home video being recorded now? Who is taking this video? Who sees it? This is the seemingly sticky part of the free house cleaning service.
Shift says that “any personal in it is anonymized before the recording is edited,” but clearly, removing “any personal” is at least one kind of editing. All that said, clearly, thousands of people are quite willing to trade their personal data for free house cleaning.
The big revelation here, however, is how valuable bot training data becomes. Cleaning the house can easily cost anywhere from $50 to $250, depending on what the cleaners do, and somehow that data is valuable enough to robot makers like Figure, 1X, Agibot, Apptronik, Neura Robotics, the UK’s Humanoid and others to make it worth it.
“The home environment is the biggest challenge,” said Mat Gilbert, director of AI and data at Synapse. he told me on a TechFirst podcast last year. “It’s pretty much the final frontier for autonomous robotics. It’s very unstructured environments.”
Gilbert says the first wave of physical AI happened largely out of public view, inside warehouses and factories, but the next wave will be visible to everyone, with service robots appearing in airports and restaurants, bringing us drinks and food. And, I might add, in our homes.
This is one place American humanoid robot maker 1X is explicitly targeting with its Neo robot. i just interviewed 1X product manager Dar Sleeperand will post it soon.
Data is the key to making robots smarter, and it looks like Shift has found a way to make us give up our data very happily.
One person I do think about though is the real person who does the cleaning just to train the humanoid robots to do the same job. This is counterintuitive, and it should be: we want robots to do work so that humans can do and be more. But we also want people to be able to support themselves and their families, and right now we’re doing that by trading time for money.
We are on the verge of breaking this contract and have no obvious news.
