The Sexism of Algorithms, Under Review in the Chilean Desert

French Journalist Judith Duportail discovered in a research that Tinder's algorithm rewarded men with higher levels of education by showing their profile to more female users, while women with the same level of education were penalized, reducing their apparences on the dating app. The example described in the book The Algorithm of Love: Trip to the Bowels of Tinder (Contra publishing, 2019) was one of those cited this Saturday by the Doctor in Mathematics Andrea Vera in the talk Is the Algorithm Sexist? Gender and bias in science, as a part of the program of one of the largest science outreach festivals in Latin America, Puerto de Ideas. The meeting took place in Antofagasta — in the middle of the world's driest desert — about 1,400 kilometers north of Santiago, Chile.

In the talk moderated by the Journalist Paula Escobar, Vera and Verónica Undurraga, history Ph.D., they unraveled how women's access to knowledge has been hindered for centuries and how social biases are shaping algorithms. "We are using algorithms to delegate fundamental decisions that affect people's quality of life. For example, whether or not they are eligible for a bank loan, qualified for a job or, in the case of the United States, whether a defendant is eligible for a bail. If we are delegating these decisions, we are concerned that they may cause racist and sexist logics", warned Vera on the 10th anniversary of the festival that gathers experts from different countries and that, year after year, becomes a cultural festival, both in the meeting in northern Chile and in Valparaíso, which takes place in the second half of the year.

The algorithm, by feeding on data provided by humans, constructs a world based on a view that is not "neutral and objective", said the mathematician. If it is asked to calculate how much a company's employees should earn, it will show that men's salaries are higher than women's, because the pay gap figures will be its benchmark. To change this bias, Vera postulated that scientific communities should diversify, and the cultural statute given to certain types of knowledge should be reviewed. "ChatGPT can be a source of fake news (...) That characteristic of truth that we sometimes associate with scientific knowledge or whatever comes from technology is particularly dangerous", she added.

To illustrate what happens in society and, therefore, in the construction of algorithms, Undurraga used an image of Río de la Plata, which appears to move easily, but carries an enormous amount of sediments. "Women go to university, they have the right to vote, they are gradually entering engineering and mathematics careers, but there is still an undercurrent that makes it difficult and considers certain knowledge as appropriate and inadequate," she said. "All technology help desks are women. That is where these gender stereotypes are reproduced. Some women in power continue there because they feel out of place", said the historian, who emphasized the importance of generating professional networks to implement gender equity policy and to prevent discrimination and abuse.

Women's barriers to breaking the glass ceiling extend to such basic Internet issues as image banks. And it has always been like that. If in 2015 someone searched on Google for the word "CEO" — the acronym for male or female executive of a company — in the first 14 results appeared only images of men, as the University of Washington study “Unequal Representation and Gender Stereotypes in Image Search Results for Occupations” reported. Only 11% of the first 100 images were women, while in real life 27% were female executives according to the U.S. Bureau of Labor.

In a review of the difficulties of Western women in the world of knowledge, Undurraga recapped key moments that have portrayed the marginalization of women from the 17th century, when Telemachus expelled his mother Penelope from the public space, through Eva magazine, which in 1949 described female doctors caring for the corpse "as if it were a baby", until now, where studies show that teachers ask more difficult questions to male students than to female students. For the latter reason, progress towards non-sexist education was one of the central themes mentioned by the researchers in order to destroy the stereotypes entrenched in Western culture and thus fight the biases of algorithms.

Undurraga emphasized the importance of understanding that gender inequality affects all of society. On the one hand, it inhibits women from entering science, technology, engineering, and mathematics fields (STEM careers), but also men from entering careers associated with women, such as teaching and nursing. "Talent has been lost to both sides", she remarked. On the other hand, it has economic consequences. Women tend to choose less lucrative careers, which hampers the household's livelihood. A female student with a score similar to that of a male student in the university entrance exam is 25% less likely to apply to civil engineering and 12% less likely to apply to technology, according to a study conducted by researchers at the Institute of Advanced Studies in Education of the Universidad de Chile and the Universidad Alberto Hurtado, published in 2020.

Article from El País AMERICA