“Just Like, Risking Your Life Here”: Envisioning AI As a Supportive Agent to Bolster Women’s Strategies for Assessing Risk of Harm in Online Dating

The upsurge of online dating usage has opened additional avenues for infliction of harm, especially against women. This study explores how women’s interactions with users and AI contributes and can contribute to respectively ameliorating decision making when choosing whether to initiate communication and whether to convert online communication to offline based on the degree of risk of harm may pose. We evaluate how women envision AI supporting them in assessing risk of harm in online dating, and how this risk assessment AI could influence their current communicative practices. The research, thus, presents an interview study in which twenty cisgender and transgender women participated in focus groups and one-on-one interview sessions, later open coded for qualitative analysis. Women produced AI models augmenting their strategies for assessing risk of harm with communicating through online modalities (textual and otherwise) or meeting given users offline. Contributions include conceptualization of a risk assessment AI that works alongside women instead of replacing their role in risk assessment completely, anticipated features to inform the risk assessment AI model are not only vast but also interdependent in the decision-making process, demonstrated need for user-personalization of risk assessment AI models, and provides methodological suggestions for future work in participatory design of AI.