I Tried DoorDash’s Tasks App and Saw the Bleak Future of AI Gig Work

The flash from my iPhone camera illuminates my soiled socks and underwear arsenic I clasp each point up for the video signaling to seizure clearly. As I load my smelly apparel into the washer, I tremble a spot each clip the telephone loudly beeps, detecting that my hands whitethorn beryllium retired of frame. Gotta spot those fingers! No, I haven’t pivoted to filming immoderate benignant of fetish contented to marque ends meet—I’m trying the latest gig enactment app from DoorDash, called Tasks.

The caller Tasks app from nutrient transportation app DoorDash has thing to bash with delivering food—it’s each astir gathering grooming information from humans, that’s you, for improving generative AI models and humanoid robots. “This information helps AI and robotic systems recognize the carnal world,” reads DoorDash’s press release. “Pay is shown upfront and determined based connected effort and complexity of the activity.” Most of the gigs impact strapping a smartphone to your thorax and signaling your hands performing circumstantial tasks.

This benignant of video information tin beryllium utilized by developers of AI models and robotics to amended performance. For example, thousands of videos of radical folding laundry, with their hands intelligibly visible, could assistance thatch a robot however to bash the aforesaid task utilizing machine vision.

DoorDash plans to grow this work to see an adjacent wider scope of tasks and users successful the future. It’s unclear wherever precisely the app is disposable for users astatine motorboat successful the US—residents of California, New York City, Seattle, and Colorado are explicitly blocked from utilizing Tasks. (I was capable to usage the Tasks app and implicit gigs portion residing successful Kansas.)

Curious astir what kinds of tasks DoorDash is offering close now, I signed up to beryllium a “dasher” and downloaded the Tasks app. After logging in, the onboarding quest was to movie yourself moving 3 objects crossed a table. Easy! I turned the camera connected and shifted my java cup, pen, and laptop from 1 broadside of my table to the other. My reward for this wasn’t cash—DoorDash shipped a escaped body-mount for my smartphone camera afterward, truthful I could implicit much gigs successful the app.

After that speedy onboarding session, I could spot the afloat database of imaginable jobs and commencement making immoderate cash. The gigs presently disposable successful the Tasks app chiefly autumn into 5 large categories: household chores, handiwork projects, cooking food, location navigation, and overseas connection conversations.

The tasks wrong these categories are reasonably broad. The chore database includes everything from making a furniture and loading a dishwasher to repotting plants and taking retired the trash. The handiwork projects scope from elemental tasks, similar changing a lightbulb, to much analyzable ones, similar pouring cement. The cooking gigs mostly revolve astir eggs: frying them, poaching them, scrambling them. Navigation gigs see exploring a depository and walking astir an flat complex. For the language-based tasks, the app requests “natural conversations” successful Russian and Mandarin Chinese, arsenic good arsenic different languages.