DoorDash Is Using 8 Million Delivery Couriers' Labor to Train AI and Robotics Systems via Its 'Tasks' App

In March 2025, DoorDash launched a separate app called 'Tasks' that pays its 8 million delivery couriers to perform micro-tasks designed to train artificial intelligence and robotic systems, such as filming themselves washing dishes, photographing restaurant menus, or recording speech in foreign languages. Uber followed in October 2025 with a similar program through its AI Solutions Group. Workers are generating training data that could ultimately be used to automate their own jobs, with no ownership stake in the resulting AI models or intellectual property. Why it matters: Gig workers desperate for additional income are generating the training data that will power the autonomous delivery systems designed to replace them, so the platforms acquire AI training data at gig-worker pay rates far below what specialized data labeling companies charge, so the resulting AI and robotics systems will eventually reduce demand for human delivery workers, so the workers who built the training datasets will be displaced without severance, retraining, or any share of the value their data created, so gig platforms achieve a uniquely exploitative cycle where the workforce funds its own obsolescence. The structural root cause is that gig workers as independent contractors have no intellectual property rights over the data they generate during platform work, no collective bargaining power to negotiate terms around automation, and no legal framework that requires platforms to share the economic gains from AI systems trained on worker-generated data.