On May 21st, 2025, the Feminist AI Network hosted a spotlight discussion as part of the exhibition “Her mit den Portr[AI]ts!” at the University Museum Bonn. The event centered on questions of gender bias in AI and the role of feminist perspectives in shaping more just technological futures.

At the heart of the exhibition was the collection “Versäumte Bilder” by artist Gesine Born, who used artificial intelligence to generate portraits of women whose scientific contributions have historically been overlooked. A unique feature of the exhibition was the presentation of “AI outtakes”; images that were discarded during the creative process. These outtakes offered a rare look behind the scenes and raised important questions: What influences our choices of what is shown? And what remains invisible and why?

This promted discussion about the highly aestheticized representations produced by the AI. Most of the portraits presented the women as conventionally attractive, youthful, and smooth-skinned – reflecting dominant societal beauty norms. It was pointed out that for the AI women seemed to fall into two categories: young or old, with the age range in between (often associated with professional maturity) rendered invisible. This absence was recognized as a reflection of broader cultural biases, and several attendees shared how this erasure resonated with their own experiences.

Indeed, the exhibition prompted responses from contemporary women scientists, several of whom are pioneers in their fields. They spoke out about the ongoing invisibility of women’s work in science today, echoing the very gaps the exhibition seeks to address. Their stories and contributions became acts of visibility themselves. 

Building on these reflections, the spotlight discussion titled “Invisible Women – Gender Bias and the Future of Feminist AI” featured a talk by Eva Maria Hille and Dr. Julia Maria Mönig. Their presentation examined bias in datasets and algorithmic decision-making, while also engaging with the role of AI-generated fakes, the politics of representation, and the implications of gendered digital violence.

A key theme that emerged was the ambivalence of visibility: While increased representation is often seen as a remedy for exclusion, it also raises critical questions about what is being made visible, how, and for whom. More visibility can sometimes reproduce existing stereotypes or create new modes of erasure. AI-generated images, even when known to be fake, still shape perceptions of gender, power, and embodiment in lasting ways.

The open discussion allowed for deeper engagement, critical exchange, and reflection. But still: Between visibility and distortion, recognition and reproduction, historical absence and technological design, much remains to be negotiated.

We invite you to take part in this ongoing conversation and to join us at the next event of the Feminist AI Network.