detrans.ai: A Counter Narrative | Peter James Steven
Peter James Steven built detrans.ai to give detransitioners a voice that mainstream AI systems routinely silence. For European audiences — from Sweden's clinic closures to the Netherlands' policy reversals and the EU's emerging AI regulation — the episode raises an urgent question: when young people turn to AI for answers about gender identity, what narrative do they encounter, and who decided that?
Peter James Steven is not a clinician or a researcher in the traditional sense, but a technologist who found himself confronted with questions about gender medicine that the dominant culture seemed unwilling to ask. His response was practical and precise: he built detrans.ai, a chatbot trained on the documented experiences of detransitioners, designed to offer the kind of information that users will not readily find when they query mainstream AI tools such as ChatGPT. The Beyond Gender hosts explore both the architecture of that project and the convictions behind it. The technical conversation is genuinely illuminating. Steven walks through the model selection process, the construction of the underlying database, and the challenges involved in building a tool that is both accurate and ethically defensible. The statistics the database has accumulated — covering transition and detransition patterns — give the episode an empirical texture that goes beyond anecdote. What emerges is a picture of a population whose experiences have been systematically underrepresented, not only in clinical literature but in the AI systems that millions of people now consult for health information. For a European audience, this matters in ways that are immediate and institutional. Across the continent, countries that once led the expansion of youth gender medicine are now in retreat: Sweden, Finland and Denmark have restricted paediatric hormonal treatment on precautionary grounds; the Netherlands, birthplace of the so-called Dutch protocol, is conducting its own review. Yet the AI tools that European teenagers and parents reach for every day still largely reflect an affirmation consensus that predates these policy reversals. The EU AI Act obliges providers of high-risk AI systems to demonstrate accuracy and non-discrimination — but the question of whose experiences count as data worth including has not been resolved by regulation alone. Steven also addresses something subtler: the role of pronoun use and language in shaping how people understand their own histories. This resonates across Europe, where several countries are actively debating whether affirmative language in clinical and educational settings forecloses rather than opens psychological exploration. The argument detrans.ai embodies — that a counter-narrative is not a harmful narrative but a necessary one — maps directly onto conversations happening in gender clinics and therapy rooms from Amsterdam to Berlin. The episode closes with personal reflection on identity and sexuality, a reminder that behind the data and the code there are individual lives. For European clinicians, policymakers and family members looking for resources that take detransitioner experience seriously, the conversation points toward a tool that at least tries to hold space for complexity. Whether AI can do that reliably remains an open question, but the fact that someone has built a functioning alternative is itself significant.
