Carpool algorithms challenged as black woman quotes more than white man for same trip to Toronto
TORONTO – A black woman and a white man have been cited at drastically different rates on the same car trip, leading to calls for more transparency and questions for tech companies about whether they are monitoring correctly the computer programs that determine the prices.
Joanne Parks and Adam Ferriera said they put identical departures and destinations in their Uber apps in Toronto last week and were surprised when Parks’ fare was more than $ 25 higher.
“I said, oh, that’s what my Uber is, my friend admitted the fare was different with him, and he said, ‘That’s a little weird,’ Parks told CTV News. Toronto She said they tried again, within seconds, and got the same result.
Ferreira was offered a trip from Dupont Street and Christie Street to Rebel in the Port Lands for $ 32.58. Parks was offered the same trip for $ 57.80, or 77% more.
“It was surprising, I was a little shocked,” Parks recalls.
The couple took a photo of them holding the two phones next to each other with the caption: “Just tell me the reason is not what I think @Uber” – she was retweeted over 25,000 times.
Ferreira said he just wanted to know what was behind the different treatment.
“It could be racial, it could be that she is female and I am male and women are more dependent on cars, especially at night, so they can charge women more. It could have been anything, ”he said.
Uber pointed out that one phone shows “fares are higher due to increased demand” and the other shows “fares are much higher due to increased demand,” claiming that its algorithm uses a “surge pricing” where prices change due to a variety of factors.
Uber told CTV News Toronto in a statement, “Surge prices change in real time and have nothing to do with the identity of the passenger. Uber is committed to being an anti-racist company and has no room for discrimination. “
Unlike taxis, there are no regulated fares for ridesharing companies and their algorithms constantly change prices based on many factors. Drivers say they don’t even know what is being charged.
Ronald Baecker, professor of computer science at the University of Toronto, said some algorithms are not programmed, they are trained with real world data. This data can be biased and this bias can be built into the application without programmers realizing it.
“These algorithms treat people of color differently from white people, treat women differently from men,” he said.
These programs are often black boxes, forming themselves, without any input from the companies that run them, he said.
“My guess is that no one at Uber really understands why this decision was made,” he said.
A study in Chicago showed that carpooling algorithms offered higher prices to people of color, the poor, young and highly educated. This study was possible because Chicago required companies to disclose their prices, which Toronto does not, he said.
Baecker says Toronto should follow Chicago’s lead and disclose its prices so that all trends are clearly visible.
And he said the federal government should consider adopting regulations similar to Europe’s, which require that if a person wants to know the reason behind an algorithm that made a decision about them, they be notified.
These ideas were supported by Parks and Ferriera, who say the experience changed their perception of their ridesharing apps.
“Whenever I use the app now, I will always compare it to the guy next to me to see if I get a better rate,” Ferreira said.
Parks said she deleted the app and will no longer be using it.
One option for her – to return to taxis, which have apps and a regulated fare. For the trip they tried, the fare would be around $ 25.60 no matter who hails it.