AI and Pride, the conversation nobody is having

We’re told AI is neutral — objective, just maths. It isn’t. Someone went and tested that, and found it treats LGBTQ+ people worse, measurably, again and again — not because anyone typed the bias in on purpose, but because it learned from us and copied what it found. This episode is about what that looks like in practice: whether LGBTQ+ bias in AI is real, where it comes from, deadnaming and rigid binary assumptions, what happens when a chatbot starts agreeing with you, and why the same thing that breaks it could also fix it. With Kaaren Hilsen, who runs the Telenor AI Factory, and Sophie Willums, a psychologist and summer intern who quietly took apart everything Kaaren thought she knew. There’s a benchmark in here that ranks AI models from ‘slightly concerning’ to ‘gravely concerning’ — and yes, that’s a real scale.

Featuring Kaaren Hilsen (she/her, Telenor), who hosts this episode, with Sophie Madeline Willums (Telenor). Introduced by Yuki Morais (they/them, Nokia).