Lutter contre les stéréotypes
One crucial step toward ethical AI is to fight against stereotypes both in the AI workplace, and its products algorithms and data. The Diversity Commission addresses this major challenge.
An algorithm can become "sexist" due to the fact that learning data can be biased. In addition, sexist biases can creep into algorithms, as they are 88% designed by men*. In response to this, biases in data and algorithms need to be detected and eliminated, which is a major research focus of ANITI. But another, equally important solution is to increase the presence of women researchers and engineers in computer science.
Historically, women occupied an important place in computer science, but they are increasingly absent from this field. Moreover, AI is multidisciplinary, from mathematics to cognitive sciences, neurosciences and computer science. Gradually all of these subjects have become more masculinized. The topic Digital studies and Computer Science specialty was chosen in 2020 by 15 to 20% of high school students, of which less than 3% are girls. There are barely 10% female students at the university in computer science.
There are barely 10% female students at the university in computer science. In the end, AI suffers from this lack of female participation**. It is, however, possible to re-interest girls in science in general, and digital in particular. For example, schools such as Girls who Code in the United Statesare very successful.
Given this observation, ANITI has set up a commission to contribute to gender diversity in AI. Beyond improving the rate of female presence in ANITI, the aim is to sensitize the general public to the issues of ethical AI and to encourage women of all ages to work in AI fields. * Aude Bernheim and Flora Vincent, Artificial Intelligence: not without them, collection "Equal to Equal",
* Aude Bernheim et Flora Vincent, “Intelligence Artificielle : pas sans elles“, collection « Égale à Égal »
** Science, technology, engineering and mathematics.
To read: Plus de femmes dans les STEM : science, technologie, ingénierie et mathématiques. ( journaldunet.com)
The commission initiates and launches actions to combat prejudice about AI and to fight against gender stereotypes. The target is primarily girls and boys in the school environment, because it is known that stereotypes are entrenched from an early age. The commission also aims to increase over time the number of girls in AI professions.
The commission ensures the representation of women in the events in which ANITI participates. It is also vigilant on the gender aspect in the diffusion of scientific culture, such as the creation of an educational game for high school students that includes awareness of bias and a debate on the challenges of an acceptable AI.
These actions are carried out in coordination with the Rectorate, the Region, academic institutions and associations working against inequalities between men and women. In addition, a network of ambassadors 'Mixity in AI' from each partner company of ANITI will be set up and will serve as a relay and force of proposals for all these actions.
Composition of the committee
- Marjorie Allain-Moulet, Industrial Co-Pilot of the IA research program Acceptable
- Catherine Dematteïs, communication manager at the Occitanie Ouest Delegation of the CNRS
- Mélanie Ducolombier, communication manager, Université fédérale Toulouse Midi-Pyrénées
- Catherine Gadon, responsible for the dissemination of Scientific and Technical Culture, Université fédérale Toulouse Midi-Pyrénées
- Corinne Joffre, Secretary General of ANITI, Université fédérale Toulouse Midi-Pyrénées
- Gwenael Kaminski, project manager, dissemination service for Scientific and Technical Culture, Université fédérale Toulouse Midi-Pyrénées
- Florence Sedes, IRIT, President of the group « Femmes et Informatique » of the Société Informatique de France; and Women and Science Occitanie Toulouse
- Chantal Soulé-Dupuy, Director of the ANITI Graduate School
- Nicolas Viallet, operational director of ANITI, Université fédérale Toulouse Midi-Pyrénées