Uber said in October 2025 it was piloting digital tasks in the Driver app, including photo uploads to help train AI models. The tasks had been tested in India and were launching in the U.S. The interesting part is not the size of the data-labeling market. It is whether a driver can make worthwhile money from a task between rides.
I had written that a driver could stack a few one-dollar jobs into $5–10 of downtime income. I had no completion-time, rejection-rate, or task-volume data to support that. A job that pays a dollar and takes six minutes, plus a switch back to driving, may be a poor use of a driver’s attention. It may still be useful in a quiet hour. Uber should show the pay per completed task and how long people actually spend getting paid.
The platform advantage is real: Uber already has people opening the app for work and a payment system that can handle small jobs. Uber AI Solutions also sells data work to clients. But labeling quality needs review, and those checks cost money. Existing distribution does not make the new product free to run.
I would watch whether drivers return to the tasks after trying them, whether clients buy repeat work, and whether ride availability suffers when drivers switch contexts. If those numbers are good, Uber found a useful second kind of work. Until then, “monetizing downtime” is a pitch, not earnings for the person waiting in the car.