DoorDash’s Tasks App Shows How AI Is Reshaping Gig Work Into a Bleak New Reality

DoorDash’s new “Tasks” platform marks a stark pivot from its core food‑delivery business to the burgeoning market for human‑generated training data, a shift that analysts see as an attempt to monetize the company’s massive gig workforce amid slowing restaurant margins. According to a Wired test of the app, workers are asked to strap a smartphone to their chest and record mundane activities—folding laundry, changing lightbulbs, or frying eggs—so that generative‑AI models and future humanoid robots can learn to recognize and replicate those motions. The company’s press release frames the data as essential for “AI and robotic systems [to] understand the physical world,” and promises “pay… determined based on effort and complexity,” but the initial compensation is often non‑cash, such as a free body‑mount for the phone, rather than a wage.

The catalog of tasks currently spans five broad categories: household chores, handiwork projects, cooking, navigation, and foreign‑language conversation. Wired’s reporter noted that even simple chores like loading a dishwasher or repotting a plant are listed alongside more technical gigs such as pouring cement or conducting “natural conversations” in Mandarin and Russian. The app’s onboarding exercise—moving three objects across a desk—serves primarily as a data‑capture test, after which workers can select higher‑pay gigs. However, the platform is geographically restricted; residents of California, New York City, Seattle and Colorado are blocked from participation, a limitation that Wired attributes to regulatory concerns, while the author was able to work from Kansas.

CNET’s coverage underscores DoorDash’s positioning of Tasks as a supplemental income stream for its drivers, noting that the company markets the service as a way to “earn more money” while simultaneously feeding the data pipelines of AI developers. The article does not disclose exact payout rates, but the emphasis on “pay you to train AI” suggests that DoorDash is betting on the willingness of gig workers to accept lower, task‑specific remuneration in exchange for flexible, on‑demand work. This model mirrors earlier crowdsourcing efforts by firms like Amazon Mechanical Turk, yet DoorDash’s scale and brand recognition could amplify the reach of such data‑harvesting operations, potentially normalizing a new class of low‑skill, high‑volume AI‑training gigs.

Industry observers warn that the shift raises labor‑rights questions, especially as the tasks involve personal spaces and potentially sensitive content. The Wired piece describes the experience of filming one’s own laundry, complete with “flash from my iPhone camera” and audible beeps warning when hands drift out of frame, highlighting the invasive nature of the work. Because the data is intended for training vision systems, the recordings must capture clear, unobstructed views of hands and objects, effectively turning private household activities into publicly exploitable datasets. While DoorDash claims transparency by showing pay upfront, the lack of cash compensation for onboarding and the reliance on equipment giveaways could blur the line between gig work and unpaid data collection.

From a strategic standpoint, DoorDash’s foray into AI data acquisition reflects a broader trend among platform companies to diversify revenue streams beyond their legacy services. As the AI arms race intensifies, the demand for high‑quality, annotated video data is soaring, and DoorDash appears to be leveraging its existing driver network to meet that demand. The move also signals a potential redefinition of gig work: tasks that once centered on delivering tangible goods are now being repurposed as “micro‑labor” for machine learning pipelines. Whether this model proves sustainable will depend on how workers weigh the modest, often non‑monetary rewards against the privacy implications and the long‑term value of their contributions to AI systems.