We may receive a commission on purchases made from links.
Uber is testing a new way to keep its U.S. drivers busy and earning money. But, oddly enough, it has nothing to do with driving people around or delivering meals to doorsteps. Through a new pilot program Uber’s calling “digital tasks,” drivers can complete small online jobs like recording short voice clips or uploading photos to help train the company’s artificial intelligence models. And it’s all in service of Uber’s AI Solutions division.
This pilot program is something Uber wants its drivers to participate in during downtime, no matter if that’s on or off the clock. It’s already gone through a successful beta test in India, and Uber now plans to expand the trial across the United States before the end of 2025. To take part in it, drivers will have to opt in through the Driver app’s Work Hub. That’s where they’ll find Uber’s “Opportunity Center,” which is where these digital tasks will live. Once a task is completed, payment gets credited to the driver’s balance within 24 hours.
What will Uber Digital Tasks look like?
Example screenshots from Uber’s recent launch presentation don’t give a ton of hints about what potential rates for these digital tasks could be. Visual aids seen during the launch presentation suggested jobs could pay anywhere from 50 cents to a dollar for a two to three minute task. That said, pay will ultimately depend on things like complexity time commitment on a task-by-task level. The number of assignments will vary, too, depending on what Uber’s clients’ needs are. (That means tasks could be few and far between during slower periods).
Examples of the specific types of tasks users should expect to see, include recording yourself speaking in your native language, submitting written documents, and uploading specific images from your everyday life. So, it’s really just your average, run-of-the-mill AI model training materials. But, instead of pulling from the internet or other data sets, Uber’s pulling from its user base. Because we know that Uber’s AI Solutions division gives audio, video, image, and text datasets for businesses developing its own machine learning systems, we can assume the stuff drivers submit will go toward that database.
What this says about the data labeling industry at large
Uber’s foray into the AI training space isn’t hard to understand. The company has a unique opportunity to leverage its existing network of millions of drivers to create a ready-made global workforce for crowdsourced AI training. How many other data labeling companies in the history of AI can say the same?
For context, the global data labeling industry is utterly exploding in value right now. Estimates back in 2018 figured the industry would surpass a billion dollars by 2023, but by 2024, it was valued at nearly $4 billion. Updated forecasts see it exceeding $17 billion by 2030. Meta continues to expand its AI data centers, as well, including paying around $14 billion for their own labeling firm.
It does raise questions about data annotators as a whole, though. Some are earning as little as $2.50 an hour without benefits (depending on their region and types of tasks being completed). Meanwhile, the AI companies they work for approach trillion-dollar valuations. For Uber drivers, the irony is probably not lost: insufficient pay is a huge pain point for drivers, and now their employer is kicking a couple bucks their way and calling it an “opportunity.”
