Possibilities of Artificial Intelligence (AI) for the prevention of gender-based violence

Abstract

: Introduction and aims: The massive emergence of artificial intelligence (AI) tools presents a significant challenge for feminist research: their gender biases to the detriment of women. This study explores the potential of AI as a tool for detecting gender-based violence in films and series through film analysis by means of the Bechdel-Wallace test. Methodology: A self-assessment questionnaire was designed, based on the Bechdel-Wallace test and other models that analyze the representation of women in audiovisual products. The generative AI Copilot (v.2024) was used to design a questionnaire, and its results were compared with those of a group of 29 university students. The questionnaire developed by Copilot (v.2024) aligned with the prevention and awareness principles established by the research team, demonstrating that AI, when guided and supervised, can generate useful tools for the critical analysis of gender representation. The students applied the questionnaire to cultural products of their choice, maintaining a critical distance that facilitated the identification of gender-based violence and reflection on gender representation in the media. Results: The responses from the pilot group were compared with those generated by Gemini 1.5, yielding a 78% coincidence between both. However, in 13% of the cases, human responses showed a deeper and more nuanced analysis than those of the AI, which tended to provide standardized answers without fully capturing the complexity of certain power dynamics. Implications: The study suggests that generative AI can be a supportive tool in educational contexts and the prevention of gender-based violence, provided its use is supervised and complemented with critical and reflective human analysis, grounded in feminist theory and research.