Examples of the reproduction of sexist and racist discrimination by artificial intelligence have multiplied in recent years. This is not surprising, as the algorithms that allow AIs to function are fed by real data sets. They reproduce the biases of our societies and even have an unfortunate tendency to run in a loop, and thus reinforce them.
The number of women in the AI professions is particularly low and this has played an important role in the fact that women are thus under-represented in algorithms. According to Syntec numérique, women represent only 27% of employees in the sector
and only 11% in cybersecurity. It’s even worse in AI, with 10% of employees working on the subject at Google and 15% at Facebook’s specialist lab, according to a Wired survey.
A concrete example of the
problems in these algorithms: Last October, Amazon, which had been using artificial intelligence since 2014 to automatically sort CVs, ended this selection system, Reuters reported. Based on data collected between 2004 and 2014, the AI systematically gave low marks to applications from women for technical jobs, such as web developers. The main reason: during this period, the company hired almost exclusively men. The AI was simply reproducing what was being done, including sexism.
While researching the subject, a work by Cathy O’Neil particularly caught my attention.
She is a highly placed mathematician in the United States who has worked in large companies and banks. In 2016, she published a book ‘Algorithms Weapons of Math Destruction’. Here she raises a civic alarm: we cannot remain spectators of a world where we are increasingly dependent on opaquely designed tools, used for commercial or security purposes and resulting in exacerbating inequalities.
In my opinion, it will be very difficult to change the system because algorithms are now established in almost everything we do. However, it is necessary to rethink the way they are made in order to be able to change things because the consequences of these technological tools are not negligible, and we don’t know yet how far it could go. Raises of awareness such as what Cathy O’Neil denounces in her book might be very helpful, can make a significant difference. We need to actively seek out the flaws in these programs in order to improve them.
BIBLIOGRAPHY
Fouquet, C., 2018. Syntec Numérique se mobilise pour les femmes !. [online] Syntec-numerique.fr. Available at: <https://syntec-numerique.fr/actu-informatique/syntec-numerique-se-mobilise-pour-femmes>
Myers, B., 2019. Women and Minorities in Tech, By the Numbers. [online] Wired. Available at: <https://www.wired.com/story/computer-science-graduates-diversity/>