Female Models in AI and the Fight Against COVID-19

Claudia Guerrero, Santiago MazuelasF1000Research 2022 · DOI

TL;DR

This opinion piece highlights the persistent gender imbalance in Artificial Intelligence (AI) and argues for increased visibility and recognition of women researchers to attract and retain talent. It examines the causes of this imbalance, including the lack of role models, and proposes solutions like fostering collaboration and promoting active participation of women in conferences, using the AI4FA COVID-19 workshop as a case study.

What Problem Does This Paper Address?

The paper addresses the critical issue of gender imbalance in STEM fields, with a particular focus on Artificial Intelligence (AI). Despite progress, women remain underrepresented in AI, impacting innovation and the future direction of the field. The authors identify several contributing factors to this imbalance, including the "Marie Curie complex" (where only exceptional women are highlighted, creating unattainable role models), survivorship bias, and impostor syndrome. They argue that a lack of visible female role models discourages women from entering or staying in AI. The paper also touches upon the "leaky pipeline" phenomenon, where women disproportionately leave STEM careers, especially after becoming parents.

Methods

This paper is an opinion piece and a review of existing literature and initiatives, rather than a research paper with novel methods. The authors analyze existing data on women's representation in STEM and AI, including statistics from conferences, grant applications, and leadership positions. They also review the literature on the causes of gender imbalance and discuss the importance of initiatives that promote women in AI. A key element is the case study of the "Artificial Intelligence for the Fight Against COVID-19" (AI4FA COVID-19) workshop, which is presented as an example of a more balanced participation between genders.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • The "Marie Curie Complex" is a significant barrier: The tendency to highlight only exceptionally brilliant female scientists can create an illusion that success in STEM is unattainable for most women, leading to frustration and impostor syndrome.
  • Visibility is crucial, but not sufficient: While increasing the visibility of women in AI is important, it's not enough on its own. Addressing the structural issues that lead to the "leaky pipeline" and ensuring a supportive environment are equally vital.
  • Active participation in conferences matters: The paper emphasizes the importance of women actively participating in scientific conferences, both as speakers and attendees, to increase their visibility and foster collaboration.
  • AI4FA COVID-19 as a model: The AI4FA COVID-19 workshop is presented as a positive example of an event that achieved a more balanced gender representation, with a significant percentage of female speakers and keynote speakers.
  • Collaboration is key: The authors advocate for a collective approach to tackling gender imbalance, emphasizing the need for collaboration among all members of the scientific community, not just women.

Key Results

The paper presents several statistics to illustrate the gender gap:

  • In 2019, only 13.8% of scientific papers in AI were authored by women.
  • Only 11% of leaders in the data industry in Europe are women, dropping to 6% in Spain.
  • In 2021, only 14% of IEEE Elevated Fellows were women.
  • The AI4FA COVID-19 workshop had 33% female speakers and 100% female keynote speakers.

Limitations and Future Work

The authors acknowledge that the gradual increase of women in AI is still too slow and emphasize the need for continued efforts. They suggest that future work should focus on:

  • Raising awareness within the STEM community about the problem and its manifestations.
  • Improving the visibility and recognition of female researchers at all career levels.
  • Developing strategies to attract and retain women in AI.
  • Fostering collaboration between men and women in AI to achieve gender balance.
  • The reviewers also pointed out potential limitations, including a perceived promotional undertone for the workshop, a lack of broader geographical scope (focusing heavily on Europe), and suggestions for structural improvements and a more balanced perspective. One reviewer noted that the article did not discuss the issue of underrepresentation in AI in sufficient depth, giving more space to gender gaps in other sectors.

Summary updated

BibTeX
@article{guerrero2022female,
  author = {Claudia Guerrero and Santiago Mazuelas},
  title = {{Female Models in AI and the Fight Against COVID-19}},
  journal = {F1000Research},
  year = {2022},
  doi = {10.12688/f1000research.123599.1},
}

Loading related content...

Figure