GWU Study Analyzing Chicago Ridehailing Apps Shows Discrimination
Researchers at George Washington University posted a preprint study on arXiv.org outlining an analysis of 100 million rides by drivers for companies like Uber and Lyft that show ridehailing apps charge more to riders in black neighborhoods in Chicago, one of many examples of discrimination by algorithms.
The researchers, Akshat Pandey and Aylin Caliskan, analyzed American Community Survey data in conjunction with U.S. Census data to target minority communities for their study. ACS, an annual survey by the federal government, collects various demographic statistics about people living in the U.S. The study explains that Chicago was chosen because of a new citywide law requiring ridehailing applications to disclose fare prices, and it included neighborhood pickup and dropoff locations, making it the most complete dataset for fare pricing. As the report explains:
“The ridehailing data contains data from 100,717,117 rides in the city of Chicago from November of 2018 to December of 2019, and the taxi data contains data for 19,496,933 rides in Chicago during the same time period. Times are rounded to the nearest 15 minutes and fares are rounded to the nearest $2.50. Because Uber takes roughly 72% of the ridehailing market share in Chicago , the dataset likely is most representative of Uber’s fare-pricing algorithm. Lyft and Via take up 27% and 1% of the remaining market share .”
The study also used what’s called the “Strategic Subject List” (SSL) dataset, a list of individuals in Chicago with prior arrests, which were then given a score of being involved in a gun violence incident in the future using the Strategic Subject Algorithm, created by the Illinois Institute of Technology. It breaks down data by race, using 224,235 examples from 801 Chicago area neighborhoods.
The study found an increase in ridehailing fare prices when riders are picked up or dropped off in neighborhoods with a low percentage of individuals over 40 or a low percentage of individuals with a high school education or less. It concludes that the pricing model used by taxis—which is not altered by AI—generally leads to less bias with regard to fare pricing.
According to a story on VentureBeat.com, the researchers say their analysis shows a clear bias correlation:
“Our findings imply that using dynamic pricing can lead to biases based on the demographics of neighborhoods where ride-hailing is most popular,” Caliskan and Pandey said. “If neighborhoods with more young people use ride-hailing applications more, getting picked up or dropped off in those neighborhoods will cost more, as in our findings for the city of Chicago.”
The study also references a report published in October 2016 by the National Bureau of Economic Research that found in Boston and Seattle, male riders with African American names were three times more likely to have rides canceled and wait as much as 35% longer for rides. Another study from Northeastern University found that users standing only a few meters apart might receive greatly different fare prices.
Uber and Lyft responded to Venturebeat by email saying that they don’t condone racial discrimination. According to Uber:
“We commend studies that try to better understand the impact of dynamic pricing so as to better serve communities more equitably,” a spokesperson told VentureBeat via email. “It’s important not to equate correlation for causation, and there may be a number of relevant factors that weren’t taken into account for this analysis, such as correlations with land-use/neighborhood patterns, trip purposes, time of day, and other effects. We look forward to seeing the final results.”
Lyft pointed out that the study didn’t take into account time of day, trip purposes and other variances, but that it would consider the results in trying to create “equity in our technology.”
read more at venturebeat.com