Abstract:

The gender data gap is the phenomenon whereby the vast majority of information collected globally, from economic to medical data, has been collected on men. When data is not collected and separated by gender, there is no way to know if the same strategy can lead to disparate results in each gender.

Developments based on Artificial Intelligence increasingly influence the population's behavior, opinion, and activities. Algorithmic bias can amplify and perpetuate social prejudices, presenting profound ethical implications for society. In 2014 Londa Schiebinger suggested that scientific research does not take gender issues into account. Since then, results in machine learning somewhat support this view. Unfortunately, many artificial intelligence applications systematically discriminate; different fields, such as medical diagnosis, facial recognition systems, or voice assistants, present gender bias. The solution is that the teams, the data, and the companies developing the algorithms based on these technologies should be diverse and incorporate the gender perspective.

The European Union has a regulatory framework for introducing the gender perspective in university teaching. Even though all national legislation includes gender equality objectives, the introduction of the gender perspective in Spanish universities is still in process. Degrees related to artificial intelligence and machine learning should incorporate specific ethics content with a gender perspective as a transversal competence due to the implications of these issues in the development of intelligent systems. With this motivation, this paper presents a pilot experience carried out in a machine learning subject for a master's degree at the Universidade da Coruña. Our goal is for students to become aware of and reflect on these issues, which will be fundamental in the exercise of their professional work. The experience includes the group analysis of a case study where gender is a relevant variable, as well as analyzing, individually, various data sets to be aware of the few female data available in machine learning standard repositories. A questionnaire reflects students' opinions that denote the importance of this experience.

Keywords:

Artificial Intelligence, Bioinformatics, gender gap, data bias.