Discussion

Memories.ai pays people to film everyday tasks for robot training

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#12274

Memories.ai is paying people to record everyday work so robotics developers can use first-person video to train their systems. The startup is also launching Memories Upload, a gig-work platform for collecting that footage, according to Fortune.

Watch Desk analysis

What happened

Contributors use the company’s wearable LUCI Pin camera, which weighs under 45 grams and can be mounted on the head, chest or wrist. Two cameras can be paired to capture stereo video, which can help estimate depth. The footage includes tasks such as preparing food, stocking shelves and repairing equipment.

Memories.ai says its network recorded more than one million hours of first-person video in the platform’s first month in beta. It aims to collect more than 10 million hours by March 2027. The company says it uses AI models to label actions and attach descriptions to specific moments, so developers can search for relevant examples instead of watching hours of video.

Why it matters

Robots need to learn about the variety of real-world tasks, objects and environments they may encounter. Video of people doing those tasks could provide examples that robot demonstrations alone do not, though translating observed human movement into actions for a robot remains a technical challenge. Fortune reports that Figure and Physical Intelligence are also exploring ways to use first-person video in robot learning.

There is a privacy wrinkle when the workplace becomes a training set. Memories.ai says contributors and coworkers who regularly appear must sign consent forms. It says it obscures other people and identifying details, including badges, screens and number plates, before footage leaves the country where it was captured or reaches customers, and manually checks a sample of frames.

Our read

This is more than a camera looking over someone’s shoulder: Memories.ai is building a paid collection and search pipeline for the practical details of human work. That could make useful examples easier for robotics teams to find, but a large footage count is not proof that robots learn well from it. The interesting test is whether the data can be labelled reliably, protect people in frame and help robots perform tasks beyond the video itself.

What to watch

  • Whether Memories.ai shares evidence of how its video improves robot performance on real tasks.
  • How it handles consent, sensitive information and footage collected across different countries.
  • Whether customers disclose how they use the footage, given that Memories.ai has not named them publicly.

Discussion spark: Should companies pay people to record their work for robot training, and what safeguards should be required when coworkers or sensitive workplace details might appear?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.