Your Uber Driver Has A New Side Hustle: Training Ai For Cash

michellegibson/iStock/Getty Images Plus via Getty Images

Follow ZDNET: Add america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

• Uber launches a new, optional "digital tasks" programme successful nan US.
• Drivers tin upload contented to thief companies train AI models.
• Compensation depends connected a task's clip committedness and complexity.

There's much to being employed by Uber than simply driving astir passengers -- astatine least, that's what Uber says. Following nan description of its courier work successful 2020, Uber is now offering drivers nan opportunity to thief companies train their AI models for other cash.

How it works

According to Uber, opening later this year, drivers and couriers who opt into nan programme tin complete "digital tasks" wrong Uber's Driver app. These tasks tin see submitting a video of themselves speaking successful their autochthonal language, uploading pictures of circumstantial mundane items, aliases presenting documents written successful a different language. 

Also: This free Google AI people could toggle shape really you investigation and constitute - but enactment fast

After tasks are completed, nan net will beryllium successful nan users' equilibrium wrong 24 hours. Compensation depends connected nan clip committedness to complete tasks and their complexity.

Uber says that complete time, much integer tasks crossed a broader scope of requests will beryllium available, giving users nan opportunity to gain much money without having to thrust their car.

Also: You're reference much AI-generated contented than you think

In a blog post, Uber says this programme connects users to "companies that request existent group to thief amended their technology" -- arsenic successful AI companies that request real-life resources to train their models. Opportunities to complete integer tasks dangle connected nan request from nan companies that request them, truthful it whitethorn not beryllium a reliable root of income.

Evading copyright issues 

Calling connected mundane group to taxable photos, videos, written documents, and sound clips could beryllium a workaround for AI companies hoping to circumvent nan copyright ineligible conflict betwixt themselves and publishers. Typically, AI companies scrape training information successful nan shape of text, audio, video, images, and much disconnected nan unfastened internet; arsenic a result, grounds labels, social media companies, newspapers, independent artists, and movie studios person sued AI companies for illustration OpenAI, Suno, and Anthropic, usually claiming copyright infringement. 

Also: Is creation dead? What Sora 2 intends for your rights, creativity, and ineligible risk

Instead of training an AI image generator connected professional, credited photos of bananas, possibly nan mean personification tin thief successful speech for a fewer other dollars. It's a gig-economy rotation connected nan existing method of training AI models, successful which companies pay workers successful nan world southbound very debased wages to tag and benignant information for models to ingest. 

Privacy concerns 

It's unclear really overmuch of a trim Uber will return from participants successful this program, and nan privateness argumentation astir nan programme is besides unclear. Uber says that it will not disclose nan names aliases business goals of AI companies to participants. You tin presume that AI companies whitethorn sell, transfer, aliases clasp immoderate contented received successful nan program.