23.9 C
Munich

Study. Facebook provides biased job advertisements, gender-biased

According to a new study, Facebook shows women and men various job advertisements that may distort anti-discrimination laws.

Researchers at the University of Southern California, who studied Facebook և LinkedIn advertising delivery algorithms, found that Facebook is gender-biased, which can be legally justified by differences in job qualifications.

Men would rather see ads on Domino’s pizza delivery driver on Facebook, and women would see ads on Instacart buyers.

The trend has been maintained in higher-paying engineering jobs such as Netflix and Chipmaker Nvidia in technology companies. More women saw Netflix ads than Nvidia ads, which paralleled the gender division of each company’s workforce.

No evidence of such bias was found in the job ads submitted by LinkedIn.

The study’s author, Alexandra Korolova, an associate professor of computer science at USC, says it may be that LinkedIn is doing a better job of deliberately mitigating bias, or it may be that Facebook is simply better at taking its lead. real signals from users. gender imbalance մասին perpetuating them.

“It’s not like the user says, ‘Oh, I’m interested in this.’ “Facebook has made a decision on behalf of the user whether they are likely to be involved,” he said. “And the fact that historically a certain group was not interested in doing something does not mean that they should not have the opportunity to pursue it, especially in the category of work.”

Facebook said in a statement on Friday that it was taking significant steps to address discrimination in advertising.

“Our system takes into account many signals to advertise and test the maximum interest of the people, but we understand the concerns raised in the report,” it said.

Facebook has promised to transform its ad targeting system in 2019 as part of a legal settlement.

The social network said it would no longer allow housing, job or credit ads targeting people by age, gender or zip code. It also restricted other targeting options so that these ads would not exclude people on the basis of race, ethnicity, other categories legally protected in the United States, including ethnicity and sexual orientation.

Endless selective advertising targeting is Facebook’s bread and butter, so any restrictions placed on it can hurt the company’s revenue. Users see that the ads can be matched to more detailed details. Not just where people live, what sites they’ve been to recently, but whether they’ve been involved in the last six months or have testimonials from people who have recently bought new sneakers, even if they never wanted to. :

But even if advertisers can’t target themselves, the study shows what critics have been emphasizing for years. That Facebook’s own algorithms can discriminate, even if the work is not intentional by advertisers.

“We have not seen any public evidence that they are working on issues related to their discriminatory algorithms,” Korolova said.

Because it is not possible for every user to display every ad that targets them, Facebook software chooses what it deems appropriate. If more women are interested in certain jobs, the software learns that it should show women more of these types of ads.

LinkedIn noted that the results of the study coincide with its internal review of job advertising targeting.

“However, we recognize that systemic change takes time; we are at the beginning of a very long road,” the company said in a statement.

U.S. law allows advertisements to be targeted on the basis of qualifications rather than protected categories such as race, gender, or age. “But anti-discrimination laws are mostly based on complaints. No one can complain about being deprived of a job if they did not know it happened to them,” said Sandra Vachter, a professor of technology law at Oxford University.

“The tools we developed to prevent discrimination were human criminals,” said Wakhter, who did not participate in the USC study. “The algorithm is very different, it groups people differently, it does it in a very subtle way. “Algorithms basically discriminate behind your back.”

While Domino և and Instacart have similar job requirements for their drivers, Domino’s delivery workforce is predominantly male, while Instacart’s women are more than half. A study of driver ads in North Carolina compared to voter record demographics found that Facebook algorithms seemed to learn from those gender discrepancies and perpetuate them.

The same thing happened with Reeds Jewelers’s retail jobs, which saw more women, at the Leith Automotive dealer, which saw more men.

Researchers urge to test such algorithms more closely և to study other factors, such as racial bias. Korolova said an external audit like the USC study could do so much without access to Facebook’s proprietary algorithms, but regulators may require some independent scrutiny to check for discrimination.

“We have seen that the platforms are not very good at self-regulating their algorithms for the undesirable society, especially when their business is at stake,” he said.

LEAVE A REPLY

Please enter your comment!
Please enter your name here