DoorDash delivery drivers can now add another side gig to their rotation: training AI-powered robots.
The food delivery giant launched a standalone app for couriers, called Tasks, that allows the company’s 8 million U.S. gig workers to earn money by recording themselves doing various tasks. The data will then be used to help AI and robotics models better “understand the physical world,” according to the company’s announcement Thursday.
Tasks available on the new app include everyday chores like folding clothes, handwashing dishes and making a bed, with each gig offering a payment sum based on effort and complexity. Harder tasks, like pruning and repotting plants, offer more money.
Couriers can also get paid for recording speaking content in other languages. One listing on the app prompts Spanish speakers to have a “natural, unscripted conversation with your friends or family on everyday topics.”
“We think this will be huge for building the frontier of physical intelligence,” DoorDash cofounder and chief technology officer Andy Fang wrote in a social media post about the launch. “Look forward to seeing where this goes!”
A spokesperson for DoorDash told NBC News that the app will initially focus on activities that could help train AI or robotics, but the company plans to add other types of activities over time. The company added that the Tasks app is a small pilot compared to what’s available in the general Dasher app, where a wider array of tasks are listed for couriers to complete in between deliveries.
It’s part of a growing ecosystem of gigs that aim to farm AI training data from willing humans.
Last year, Uber piloted a similar initiative allowing its U.S. gig workers to perform additional digital tasks for money, including uploading photos and recordings used to train AI. And the data annotation industry has boomed in recent years, with dozens of platforms hiring contractors to help train AI models online.
Now, companies are increasingly seeking to capture physical data about how people move in the world. Such content is then used to help a humanoid robot, for example, learn how to load a dishwasher.
The Los Angeles Times recently reported that Instawork, a staffing app that connects businesses with local hourly workers for same-day gigs, has been recruiting workers in Los Angeles to strap on headbands with a phone mount and record themselves cleaning their homes.
Other robotics developers have developed similar data collection strategies. California-based Sunday Robotics ships a “skill capture glove” to people across the country who collect motion data by doing household tasks wearing the robotic glove. The gloves, which remember their movements, are then used to train the AI-powered home robot that the company is building.
Aside from DoorDash’s new AI training gig app, additional tasks will also roll out within the regular Dasher app for couriers. These could involve checking a restaurant’s holiday hours, taking photos of a tricky drop-off location to help delivery drivers navigate, or “giving an autonomous vehicle a hand getting back on the road,” the company’s news release stated.
“These are the kinds of real-world problems we’ve been solving for over a decade, and we realized the same capabilities that helped us could help other businesses too,” Ethan Beatty, general manager of DoorDash Tasks, said in a statement.
[/gpt3]
