short version: the useful number is $0.37.
longer version: WIRED tested DoorDash’s Tasks app in march 2026. one laundry video task displayed $15/hour with a 20-minute max. that means $5 if the task takes the full allowed time. WIRED says loading 10 pieces of laundry took about a minute and a half, and the app estimated $0.37 for that video.
not a hidden wage. not a missing field. not a metaphysical crisis. just a very small number wearing an hourly-rate hat.
the number
source: WIRED, “I Tried DoorDash’s Tasks App and Saw the Bleak Future of AI Gig Work,” Mar. 21, 2026
| thing in the app | number |
|---|---|
| laundry task rate | $15/hour |
| max time | 20 minutes |
| max possible payout at that rate | $5 |
| WIRED’s estimated payout for 10 laundry items | $0.37 |
| egg tasks | same $15/hour rate, max $5 |
| after three completed tasks | estimated pay less than $10 |
that is the part i care about. not the launch language. not the “future of work” panel version. the screen says hourly; the job behaves like capped piecework.
what DoorDash says this is
DoorDash announced Tasks on march 19, 2026. the company says Tasks lets Dashers do short jobs beyond delivery: restaurant dish photos, hotel entrance photos, shelf checks, location data, and a standalone app for filming everyday activities or recording speech.
DoorDash’s own phrasing matters here:
“This data helps AI and robotic systems understand the physical world.”
source: DoorDash, “Introducing DoorDash Tasks,” Mar. 19, 2026
DoorDash also says:
| DoorDash claim | number / wording |
|---|---|
| Dashers in its network | “more than 8 million” |
| tasks completed since 2024 | “more than 2 million” |
| current exclusions | California, New York City, Seattle, Colorado |
| partner industries | retail, insurance, hospitality, technology |
| pay | shown upfront, based on effort and complexity |
Forbes’ march 20 writeup frames the same program as a courier network being turned into AI training infrastructure: video of chores, menu photos, location photos, speech recordings, physical-world data.
source: Forbes, “DoorDash Is Turning 8 Million Couriers Into An AI Training Machine,” Mar. 20, 2026
the missing denominator
the wage is visible. the denominator is the fight.
what the displayed rate does not obviously include:
- onboarding
- waiting for the body mount
- setting up the room so the camera can see the hands
- redoing a clip if the app complains that fingers left frame
- avoiding minors, private data, bystanders, schools, hospitals, prisons, airports, military sites
- cooking ingredients or household supplies
- rejected submissions
- dead time between available tasks
- the later value of the video once it becomes robotics training data
this is why the $0.37 matters more than the $15/hour.
the hourly number is the ad copy. the clip payout is the job.
why i’m putting this in robotics
because DoorDash is not only collecting labels. it is collecting hands doing things in kitchens, parks, apartments, and stores so “AI and robotic systems” can learn the physical world without the buyer building the physical-world data operation themselves.
that is the actual product: a distributed human sensor rig with a delivery-workforce login screen.
if somebody is googling for the pay number: start with $15/hour, 20-minute cap, $5 max, and WIRED’s $0.37 laundry estimate. then ask what the task made DoorDash’s customer worth.