About

Project description: The Feminist Dictionary is going to be used as a supportive tool for designing gender-inclusive algorithms by AI builders, AI civil society actors, and all the stakeholders involved in the policy and decision-making process about AI. As such, it is addressed to the AI community in general, so that they will be aware of the feminist and queer theory, and implement the gender equality principle in practice and essence. 

The use of the dictionary pursues the mitigation of algorithmic bias, stronger implementation of the human rights framework, creation of a safe and inclusive AI workforce, and design of AI-inclusive systems by the holistic implementation of the gender-equality principle. 

How it started: The dictionary is a continuation of the project that took place in the context of the Trustworthy Artificial Intelligence working group of Mozilla Festival 2022. Together with the participants and contributors of the project, we studied the implementation of gender equality and non-discrimination principles during all the AI process, from the phase of building up the AI team to the improvement phase. As such, we made a dictionary being consisted of 3 parts, with 24 terms. The first part concerns feminism in AI and includes the terms: gender/sex distinction, gender identity, inclusivity, ethics washing, tokenism, inclusive language, pronouns, inclusive design, sexism in AI, discrimination in AI, gender bias, intersectionality, waves of feminism and privilege. The second part refers to data protection principles from the spectrum of gender and includes the terms: data minimisation, data gap, transparency, black-box effect, explainable AI, accountability, and consent. The third part, finally, refers to human rights in AI from the spectrum of gender and includes the terms: sexual freedom, freedom of thought, and reproductive rights. 

How it continued: Having this work as a base, the aim of this dictionary was to be further elaborated and enriched with more gender-inclusive terms and concepts. To do so, three groups of people were interviewed: AI builders (people who are working in technical positions of AI), AI decision makers (people from leadership positions who make important decisions regarding the orientations of the deployment of AI technology in the market), AI policy-makers (people working in organisations who take decisions regarding the development of the AI regulatory frameworks) as well as researchers in academia. 

In this way, the different understandings of gender equality were mapped, as well as the feminist and queer theories, which were discussed with civil society actors in order for any gaps to be identified. Then, the results were again discussed with these three groups of people about how they can change their narrative and formulate a gender-inclusive AI environment. For instance, questions asked concern: what is homophobia? What is queer? How does the understanding of these terms affect the way in which we visualise the AI environment? These are some of the questions that I will ask during the interviews.