Uber is testing a new program that pays its gig workers to train AI (www.businessinsider.com)

Uber is piloting "Digital Tasks" in the US, offering some Uber and Uber Eats drivers paid microtasks that directly feed AI training pipelines—things like taking photos and uploading audio clips. The program, already available in India, is part of Uber’s AI Solutions Group, which packages labeled data services for other businesses. Uber says tasks can’t be done while driving or waiting for rides for safety, and frames the effort as a way to give drivers additional earnings during downtime as autonomous vehicles loom. For the AI/ML community this signals a scaling up of real‑world, multimodal data collection via gig platforms: ready access to geographically and contextually diverse images and audio can accelerate supervised learning, fine-tuning, and domain adaptation for perception and voice models. It also raises practical and ethical considerations—data quality control, annotation standards, privacy and consent, worker pay, and bias in collected datasets—that will affect model robustness and downstream use. The move exemplifies how labeling economies are becoming integrated into platform business models, creating both a cheaper pipeline for training data and new questions about labor and regulation as automation reshapes gig work.