How Facebook increases gender inequalities in access to work

It starts in March 2022 and April 2023. The NGO Global Witness is testing the social network by publishing five ads “with deliberately neutral title and content”. The target ? Highlight posts for secretary, airline pilot, early childhood assistant, psychologist and IT facility manager positions.

Distorted algorithms

And there, surprise (or not): 94% of people targeted for the position of early childhood assistant are women, 92% for that of secretary and 80% for the position of psychologist. Instead, the job offer for the role of airline pilot was suggested by 85% of the men, that of manager of an IT structure by 68% of the men users.

Bottom Line: Facebook’s algorithms, which did all the serving and audience targeting, presented the estimated posts” with responsibility to male Internet users and women’s care-related careers. “When posting job offers on Facebook, the social network algorithm carries out a retrospective sexist and stereotyped discrimination in the target people”, hammer the three associations in a press release published on 12 June.

Recruitment through social networks, a practice that is spreading

” Today, 53% of SMEs say they use online recruitment and 82% of job seekers seek to get hired via social networks.” are suing the three associations, which also claim that the discrimination made by Meta’s algorithms is ” illegal “. The algorithms that serve these ads on social networks therefore constitute a crucial terrain in the fight for equal access to work: “Tests show Facebook algorithms retrospectively and against the will of recruiters lock women out of otherwise better-paying job opportunities”.

The three associations wish to obtain the recognition of a ” gender-based employment discrimination with the defender of rights e the potential violation of the General Data Protection Regulation (GDPR) because this text requires data processing to be fair, which is not the case for discriminatory data processing “.

Asked by our colleague from PublicationMeta, Facebook’s parent company, weakly defended itself:

“We don’t allow advertisers to target these ads based on gender.” Advertisers who are not in question here. The team adds: We continue to work with stakeholders and experts from academia, human rights groups and other disciplines on the best ways to study and address algorithmic fairness”.

However, the solution seems obvious: if our algorithms are biased, it’s because the people who design them evolve in a biased society and are also biased. And the best way to remedy this is to diversify the teams that code.

Because how do you remember Publication, a 2018 World Economic Forum report shows that only 22% of AI professionals are women. Maybe Facebook could start by showing them the job posting?

Do you like our articles? You will love our newsletters! Sign up for free on this page.