Women working in artificial intelligence (AI) warn that the lack of female presence in this field may have consequences for the development and use of these tools.
Catherine Lavoie is a doctoral student in cognitive computing at Université du Québec à Montréal (UQAM). She is currently working on the importance of integrating cultural diversity into AI.
While she is far from being the only woman conducting research in this field, she notes that her female colleagues are less present in the private sector.
According to her, women want to work in this booming field. However, she fears that it is “difficult to be taken seriously.”
In Quebec, just under 20 per cent of professional positions in AI are held by women, according to data from TECHNOCompétences dating from 2021.
In light of this observation, the Conseil du statut de la femme made recommendations in 2023 and warned of a risk of inequality between women and men in AI if this trend continued.
According to the Council, women’s participation in AI is “essential because they are more likely to take into account women’s realities and because they make up half of the society that AI is transforming significantly.”
Related:
In AI, people are required to train algorithms and manage data.
“Data is generally built by and for men,” explains Lavoie, who points out that this is not something done consciously, but it has “a direct impact on women.”
For example, if someone searches for heart attack symptoms using an AI tool, the data is more likely to relate to symptoms experienced by men, and not necessarily those experienced by women, she explains.
“It’s a lack of representation because there are no women to say, ‘Wait a minute, we took this data set, but this data set doesn’t represent us,’” she points out.
“As long as we are excluded from these discussions, there is definitely an impact, because someone who has never experienced a situation cannot take all the impacts into consideration,” adds Lavoie
Frincy Clement, head of the North American branch of the global non-profit organization Women in AI, makes the same observation.
“If you’re not represented, you don’t know about these anomalies or noises that you need to pay attention to when working with data and results, because you’re not aware of them, you don’t know,” she emphasizes.
Marie-Jean Meurs, professor of AI at UQAM, qualifies this direct link between data and the presence or absence of women in training models, pointing out that there are many parameters to consider in the functioning of AI.
However, she believes that the presence of women in the field is important in how “the entire domain” is conceived.
“We know that these tools are extremely energy-intensive and have a very negative environmental impact,” she says.
“We can say that if the entire population, and in particular the half of the population represented by women, were better represented in decision-making circles at this level, we might have a better approach to these issues,” she explains.
For Meurs, it is clear that more women are needed on boards of directors and in decision-making positions in AI if change is to be achieved.
This is the mission of the organization Women in AI, which began its activities in France in 2017 and aims to increase the participation of women in AI.
“We want women to help women launch careers in AI or advance in the field and help them develop their leadership skills,” says Clement.
She knows firsthand how important it is to create such a network, as she herself had difficulty finding a mentor when she wanted to make the leap to AI after a career in business and a bachelor’s degree in computer science.
The Canadian chapter of Women in AI was launched in 2020. The group offers conferences, job fairs, courses, mentoring, and awards.
“We are able to highlight women who work behind the scenes or behind screens and who have never been in the spotlight, so that they can tell their stories,” says Clement.
In fact, some of the women who have been recognized at Women in AI awards ceremonies have been promoted or moved into management positions, she says.
While there is still a long way to go, Clement remains hopeful that change will come for women in AI.
“Over the years, we’ve seen more and more women in this field compared to when we started six or seven years ago,” she notes.
“It used to be very difficult to find female speakers. But today, we’ve found so many extraordinary women, perhaps because they are now more visible,” she says.
However, she believes that significant systemic change is crucial to achieving this, a view shared by Meurs, who believes that the problems are more related to power and financial issues than the discipline itself.
–This report by La Presse Canadienne was translated by CityNews
