Facebook's ad system pushes stereotypes in housing, jobs, study finds
Facebook's algorithms, which match marketingmessages with viewers, leans on stereotypes when it comes to housing and jobs, according to the study by researchers from Northeastern University, University of Southern California and advocacy group Upturn.
Facebook spokesman Joe Osborne responded in a statement that the company recognizes it must do more, and said the findings would be included in ongoing discussions about changing its ad system.
The researchers advertised lumber job ads on Facebook and found that the algorithms delivered the postings to mostly white men, while secretary positions mostly went to black women.
The Facebook study also found that ads about homes for sale in North Carolina reached a mostly white audience while rental ads went to a mostly black one.
Facebook does not provide race data, according to the researchers, but they inferred it by linking general audience details to voter registration data.
Though the pictures of people in the job ads did not appear to affect the audience makeup, the photos used did appear to be a factor for Facebook's algorithm in other cases.
Showing a football or soldiers versus a flower or paint set led to a mostly male audience for an otherwise identical ad unrelated to jobs, the researchers found.
The Trump administration sued Facebook last Thursday, accusing it of selling targeted advertising that discriminated on the basis of race, in violation of the U.S.
Addressing racial and gender discrepancies in automated systems, including for facial recognition, has become a priority for Silicon Valley.
We use cookies and analyse traffic to this site. By continuing to use this site, closing this banner, or clicking "I Agree", you agree to the use of cookies. Read our privacy poplicy for more information.