Uber is moving to reduce racial bias in its services, according to a Massachusetts Institute of Technology economist whose research found a pattern of discrimination in the ride-hailing industry.
Christopher Knittel, who led a 2016 study that found Boston Uber passengers with black-sounding names were more likely to have rides cancelled than those with white-sounding names, said his team is working with Uber to investigate the issue.
Both Uber and its rival, Lyft, got in touch with Knittel and his fellow researchers, which included academics from Stanford University and the University of Washington, shortly after they published the study last fall, Knittel said.
“Both companies reached out to us very quickly and we’ve had continued conversations with them,” Knittel said in a Web event broadcast by MIT Sloan School of Management. “And we’re actually trying to design follow-up studies to minimize the amount of discrimination that’s occurring for both Uber and Lyft.”
Knittel said the team is concentrating mostly on working with Uber, with the goal of identifying and reducing racial bias in the ride-sharing service.
Uber declined to comment.
Meanwhile, Lyft said it has undertaken its own efforts to better understand the issue, including auditing “cancellation rates and quality of service in predominantly minority neighborhoods,” said spokesman Adrian Durbin. The company noted that drivers are required to agree to an antidiscrimination policy.
The findings in Knittel’s study echoed research showing the taxi cab industry has long underserved minority communities.
It also mirrored studies of another stalwart of the so-called sharing economy, home-rental service Airbnb. A study published in 2015 found that white property owners were less likely to allow users with black-sounding names to rent their homes.
Airbnb responded by creating new policies aimed at limiting discrimination, which included making it more difficult for users who reject rental request to let a different user stay in the home on the same day. The company also required users to agree to a “community commitment” not to discriminate.
Knittel said the research on Airbnb, Uber, and Lyft should cause other companies that provide services via digital technology to develop safeguards against discrimination.
“The first thing is to be aware of this,” Knittel said. “I think Uber and Lyft and Airbnb were potentially caught off-guard with the amount of racial discrimination that was taking place.” Research on those companies “gives new startups a head start” on combating the problem, he said.
The ride-hailing study also featured an experiment outside of Boston. In Seattle, researchers found that riders on Lyft with African-American names waited longer for drivers to accept ride requests than those with white-sounding names.
Lyft drivers see passenger names before they agree to pick up a ride, which is why the ride acceptance rate may indicate racial bias, the researchers wrote. Uber drivers, meanwhile, do not see passenger names until they’ve agreed to pick up the rider, meaning the cancellation rate was a better indication of bias.
The study did not examine the race of drivers, and future research should consider its role in determining bias, Knittel said. According to Uber’s own national data, as of 2014, about 40 percent of drivers were white, 19.5 percent were black, 17.7 percent Hispanic, and 16.5 percent Asian.
Knittel suggested some solutions for the companies. One option would be to no longer show drivers the names of customers requesting rides. Uber has said that providing riders and drivers with each other’s names is key to making sure riders get in the right vehicle.
Another option, Knittel said, would be to only show the names later in the process — 30 seconds before picking up the passenger, perhaps. That way, “the driver has to commit more to the ride than he or she previously had to, and that may increase the costs of discrimination,” he said.Adam Vaccaro can be reached at firstname.lastname@example.org. Follow him on Twitter at @adamtvaccaro.