Uber will offer gig work like AI data labeling to drivers while not on the road (www.cnbc.com)

Uber announced at its Only on Uber 2025 event that it will pilot a new program letting U.S. drivers and couriers earn money through small online tasks in the app while they’re off the road. The AI Solutions Group–run pilot — already trialed in India — will offer microtasks such as uploading photos to train models or recording spoken prompts in specific languages or accents; pay will vary by task complexity and estimated time, and drivers will see compensation before accepting a job. Uber says tasks won’t be tied to its autonomous-vehicle projects and won’t disclose client names or project specifics. For the AI/ML community, this signals another large-scale channel for gathering labeled data and speech samples, similar to Mechanical Turk or freelance labeling platforms. The potential upside is rapid access to geographically and demographically diverse data at scale, but it raises technical and ethical questions: data provenance and annotation quality, consent and privacy for uploaded media, biases introduced by worker demographics, and reproducibility when client/project details are undisclosed. Researchers and practitioners should watch for how Uber implements quality control, metadata capture, and privacy safeguards—because how these microtask pipelines are run will shape dataset reliability and the downstream performance and fairness of models trained on them.