Technology, contrary to what many think, is socially and culturally constructed. This means that, if we want it to provide us with higher levels of justice and equality, it is advisable to include a social and ethical dimension. In the case of Artificial Intelligence (AI), due to its own characteristics such as the use of big data and its data processing capabilities, this recommendation has become a must.
To enjoy the benefits of AI (greater effectiveness, efficiency, automation, personalization, tracking, detection, etc.), it is first necessary to ensure three things: that its use is within the limits set by the EU legal system in this matter, that the AI is robust from a technical point of view and that it is ethical. Compliance with these requirements is aimed at preventing the use of AI in Europe from leading to ethical and legal problems such as discrimination caused by biases, accountability problems, etc.
In this sense, and due to their sociocultural conditioning, one of the greatest challenges when introducing AI systems in the Spanish and European public sector is to guarantee that they do not have a negative impact on the lives of women and, ideally, that they help in the promotion of Gender equality. An issue that is often ignored and / or downplayed in discussions about AI.
But what risks can AI bring to gender equality? Although, from a theoretical point of view, gender issues are closely related to certain elements that concern AI (for example, the association of the word “intelligence” with logical and rational rationing is something historically linked to masculinity Western), from a more practical point of view there are two issues that are especially important when talking about AI and gender.
First, AI can produce gender biases in the decisions it makes if the representativeness of the data is not taken into account. The fact that the data are representative means that they reflect the diversity of social reality in the most truthful way possible. However, we now know that women are less present in certain sectors and, specifically, they have less presence on the Internet. Although the digital divide has narrowed in recent years, there are still differences: globally, 55% of men use the Internet compared to 48% of women, and in some regions, such as Africa, the imbalance of gender is even higher. Gender differences in digital skills should also be considered, which tend to leave women worse off, especially in more advanced skills. This means that the data is not only biased in favor of men, but also that, among women, those white, western and upper class, have a greater presence than the others.
Second, just as it is important to include the gender perspective in the design of public policies, personnel hiring processes, etc., it is also important in the case of algorithms. As the algorithms are systems of rules based on which the AI makes its decisions, it is necessary to introduce rules or steps that avoid gender discrimination. At the same time, the gender perspective must be present in the design of the AI. Until now, most of the chatbots and voice assistants present characteristics clearly associated with the female gender, in such a way that their interactions with users reinforce existing gender stereotypes.
Solving these problems would be the first step towards an AI that does not negatively impact women, but can we go further? Is a feminist AI that actively promotes equality between men and women possible? The answer is yes. By taking advantage of the opportunities of AI, this technology can help us promote equality in at least three areas that are essential for women.
In the fight against gender-based violence, for example, through the development of chatbots that facilitate the administrative processes to report and that provide assistance in real time to the victims. AI can also be used to predict the likelihood of recidivism of a person convicted of gender-based violence, etc. AI can be used to reduce the pay gap and economic inequality between men and women through the development of AI systems that contribute to pay transparency. In addition, AI can be used to address the specific health needs of women through the development and use of AI systems specialized in diagnosing diseases suffered by women such as cervical cancer, breast cancer, etc. .
Ultimately, only an AI with a gender perspective will allow this new group of emerging technologies to help us achieve higher levels of justice and equality. Only a feminist AI has a place in a democratic society.
Lucía Ortiz de Zárate Alcarazo She is a predoctoral researcher in Ethics and Governance of Artificial Intelligence in the Department of Political Science and International Relations of the Autonomous University of Madrid.
Ariana guevara gomez She is a pre-doctoral researcher in Gender, Technology and Public Administration in the Department of Political Science and International Relations of the Autonomous University of Madrid.