Algorithmic Gaslighting: A Qualitative Analysis of Women's Trauma Narratives in the Age of AI-Mediated Therapy

As AI develops rapidly in mental health, however, there still aren’t enough studies to show how useful these new tools can be for people who have experienced trauma due to their Gender Identity. As a result, this study highlights what women experienced when they used AI mediated therapy applications through semi-structured interviews and narrative analysis, developing an understanding of the algorithmic gaslighting phenomenon, a unique experience that occurs when the standardized response created by AI invalidates and mischaracterizes the complexity of women’s historical context concerning, GBV by providing them with generic regurgitated conceptual reframes without placing their trauma into context of their personal lived experiences. Thus, this study asserts that the current NLP models possess an insufficiency of the risks associated with using AI-mediated therapy applications in relation to traumatic experiences; therefore, these models create a substantial risk for retraumatizing victims of trauma. There must be additional feminist ethical considerations taken into account when creating the architecture for digital healthcare interventions.