detrans.ai: A Counter Narrative | Peter James Steven
Peter James Steven built detrans.ai to do what mainstream AI tools will not: treat detransitioner testimony as a coherent body of evidence. For British listeners living through the aftermath of the Cass Review and the Tavistock closure, this matters. If AI shapes what young people believe is normal or possible, a tool deliberately designed to centre detransitioner voices is a direct challenge to the information environment that surrounded NHS gender medicine for years.
Peter James Steven built detrans.ai not as a commercial product but as a personal response to a gap he saw in the information landscape. As an AI-powered chatbot trained on the accounts of people who have detransitioned, it sets out to do something that mainstream models have largely avoided: treat those voices as a coherent and significant body of evidence. In this episode of Beyond Gender, Steven walks the hosts through the project's origins, its design, and what the data it has gathered is beginning to reveal. The mechanics matter here. Steven explains how the underlying model was selected and trained, and why those choices shape what the chatbot can and cannot say. Unlike general-purpose AI tools that have been widely noted to default toward gender-affirmative responses, detrans.ai draws on a curated dataset of detransitioner accounts. The episode goes into some technical depth on this, but the underlying point is accessible: the stories encoded into an AI system determine the answers that system gives. Where detransitioner experiences are absent from training data, they will be absent from the outputs. For British listeners, this has immediate relevance. The Cass Review, published in 2024, identified a striking lack of robust evidence behind the affirmation-only model that had dominated NHS practice for more than a decade. The Gender Identity Development Service at the Tavistock was subsequently closed. Yet in online spaces, in school resources, and in AI tools routinely used by young people, the affirmation narrative remains the default. A tool designed to surface the counter-narrative — to ask what happened to people after transition and to take their answers seriously — has an obvious place in the landscape that Cass left behind. The episode also examines the statistical picture emerging from the detrans.ai database, covering trends in who transitions, why, and what eventually draws people back. Several of these patterns align with what clinicians and researchers in Britain have been documenting: elevated rates of co-occurring mental health conditions, questions about social influence, and a demographic shift toward adolescent females. The Cass Review pointed to exactly these patterns; Steven's project approaches them from the lived-experience end, through the words of people who were there. A section on pronouns and their relationship to gender identity touches on something that remains unresolved in British public life. Schools, NHS trusts, and public bodies across the UK have spent years navigating guidance on pronoun use, often without clear legal grounding and often under pressure to follow affirmation-based frameworks. The conversation here treats pronoun use not as a settled courtesy but as part of the broader set of questions about how language can shape and entrench identity over time. Steven is candid about the personal connections that motivated him to build the tool. That combination of the technical and the human gives the episode an unusual texture. What emerges is an argument for treating detransitioner testimony not as an inconvenient statistical outlier but as a form of evidence that AI, if deliberately designed to do so, can help preserve and make visible — at a moment when that visibility is harder to find than it should be.
