AI Safety

Queer characters are heroes but collective writing has a Fool bias, study finds

How fictional queer characters often shine as Heroes, yet overall writing skews toward negative tropes.

Deep Dive

A team led by Ashley Fehr used archetypometrics—a computational method that operationalizes common character archetypes like Hero, Fool, Diva, and Outcast—to analyze queer representation in fictional stories. They combined this with the Fandom LGBTQIA+ dataset to examine thousands of characters from TV, film, and literature, quantifying their placement along a straight–queer trait differential.

The results reveal a striking paradox: individual characters with the highest queer scores tend to be positive archetypes—Heroes instead of Fools, Angels instead of Demons, and Adventurers instead of Traditionalists. However, when analyzing the collective writing across all stories for the straight–queer trait itself, the researchers found a strong bias toward the Fool archetype (and away from Hero), with no meaningful loading on the other two dimensions. This suggests that while creators may write positively when consciously developing queer characters, the broader, aggregate portrayal across many-authored stories still leans into negative stereotypes. The authors caution that blindly training large language models on such corpora could perpetuate these biases.

Key Points
  • Studied 23,194 KB of character data using archetypometrics and the Fandom LGBTQIA+ dataset
  • High-queer-scored characters are disproportionately Heroes, Angels, and Adventurers
  • But across stories, the straight–queer trait shows a collective Fool bias and zero Hero loading

Why It Matters

Reveals a systemic bias in storytelling that could skew AI models trained on narrative data.

📬 Get the top 10 AI stories daily