Uber’s AI contractors face pay spikes and chaos

Uber AI – Uber’s AI training unit is recruiting PhD contractors with high hourly pay, but multiple workers describe chaotic scheduling, limited onboarding, uneven weekly hours, and sudden terminations—shaping a gig-style experience that looks far less stable than promis

A calendar invite titled “Project Sandbox Kickoff” landed in the inbox of some AI training contractors working through Uber’s AI training arm on a May evening. For at least one contractor. the timing felt wrong in the way only a long delay can—he hadn’t heard from Uber in months. even though he’d been approved to start working for the company in December.

“Project Sandbox Kickoff” was not the only sign of how unpredictable the work could be. Several contractors described shifting schedules. last-minute changes. and an onboarding gap: they said they were expected to start quickly. but received little or no training on how to vet queries or annotate data.

Uber is pushing deeper into white-collar gig work as it expands its model beyond drivers. CEO Dara Khosrowshahi said last year that Uber wants to become a “platform for work.” To reach that goal, the company has been recruiting contractors with PhDs for AI training work since late last year.

The effort also puts Uber into a competitive niche dominated by firms specializing in AI training. including Mercor and Snorkel AI. The jobs can look attractive on paper. A Guggenheim Securities analyst. Taylor Manley. told Business Insider the format can offer “better earnings potential” for AI engineers than traditional full-time employment.

Contractors said the higher pay is real. Four workers—two currently working and two who were recently let go by Uber—told Business Insider that the Uber role paid between $50 and $150 an hour. They said that compared with similar AI training contract roles they previously held. their earlier rates could be as high as $30 per hour.

For some, the trade-off was stability.

Uber’s email spelled out flexibility—and deadlines

Uber told two contractors in an email that schedules would be flexible, a key selling point for gig work. The expectation, according to that email, was to work between 20 and 40 hours each week, with management deciding how many hours each contractor would receive.

There was one major constraint: workers had to complete at least 70% of their hours between 9 a.m. and 5 p.m. PT “to facilitate real-time collaboration and regular standup meetings,” according to the email.

Google, the client, would assign each worker hours week to week. One contractor who managed a team of other Uber contractors said Google typically distributed weekly hours on Sundays for the week ahead.

But that system produced uneven outcomes. Some contractors received close to a full-time allotment. Others received zero hours and were told to check in again the following week in case hours had been assigned.

Two contractors said that workflow differed from their prior gig work at other AI training companies, where they received a consistent number of hours each week.

Onboarding, they said, didn’t match the job’s demands

Once contractors started, the experience described to Business Insider became even more stark. Contractors said they were expected to get to work right away, and that neither Uber nor Google provided training on how to vet queries or annotate data.

One worker—who previously trained AI for Google as a contractor for another company—described an orientation process that lasted days. That earlier orientation involved identity verification, signing non-disclosure agreements, and classes on how to do the work.

At the Uber gig, that contractor said the process was different: they received “no orientation, no real training on anything.”

Several contractors said they worked on projects for Google, including Project Magi, the search giant’s effort to add AI replies to search results.

Timing problems didn’t end with onboarding

The day-to-day friction described by contractors included sudden changes and delayed access to work.

One worker told Business Insider that he had been approved to start working for Uber in December but hadn’t received work until months later—right up to the point of receiving a “Project Sandbox Kickoff” calendar invite in May.

The contractor who received that surprise meeting invite said the lack of clarity continued after he started. He said at least one person on his team had been terminated since he joined, and Business Insider confirmed the termination.

He said no one at Uber or Google mentioned the termination after it happened. He found out from colleagues outside work channels. “Nothing has been discussed whatsoever,” the contractor said.

Termination can arrive before a contract runs its course

Not every contractor’s timeline followed the terms they expected.

A job description sent to contractors indicated that contracts would last three months. But some said their time with Uber and Google ended sooner.

In the fall of last year, Business Insider reported that Uber terminated some AI contractors working on a Google project about a month into their three-month contracts.

This year, a contractor who said they had experience training AI for Google through another company was terminated at their Uber gig about a week after starting. An email shared with Business Insider did not include a reason for the termination.

Another worker described a different kind of sudden stop. Uber recruited and onboarded him last fall for a job that would pay $150 an hour. but he never received any hours. In December. he said he received an email saying the project he was recruited for had ended and that Uber would reach out if more work was available.

Then, in May, Uber emailed him a new offer: record short clips of himself speaking in different environments—such as a noisy coffee shop or a quiet library—for $10 each.

The worker said he’s completed those tasks because they relate to his AI experience, even though it isn’t the job he initially signed up for. “At least it’s something for my résumé,” he said.

Uber and Google did not respond to requests for comment.

Uber AI Solutions Uber contractors AI training gig work Dara Khosrowshahi Google Project Magi Mercor Snorkel AI contractor scheduling hourly pay