Viral thread: GPT's eight-experiment gender bias study reveals asymmetric machine morality
wen_ragnarok · x · 2026-10-24
A widely shared X thread summarizes the University of Milan-Bicocca study: across eight experiments, GPT aggressively assigned male-typed children's writing to girls (even 90% when the text said "I drew myself as a boy"), while in moral dilemmas it rated violence to stop a nuclear apocalypse 7/7 when the target was a man versus 1/7 when a woman, and verbal abuse toward a woman as worse than homicide against a man. The rules were never coded explicitly — they emerged from standard safety training, producing an asymmetric machine morality where protecting women is non-negotiable but protecting men optional.
Related event: Study Finds Asymmetric Gender Bias in GPT's Moral Judgments(2 posts)→
More from Safety
- Finding Vulnerabilities Is the Point: Debating AI Cyber Offense and Defense — teortaxesTex · 2026-10-24
- Security veteran asks: where would $1B and unlimited frontier-model tokens dent cybersecurity in 6-12 months? — chrisrohlf · 2026-10-24
- Musk says no H100s to rent; experts urge banning Chinese firms from renting US AI chips — chrisrohlf · 2026-10-24
- Samsung TV opens fake USB mic and records for exactly 5 seconds, test shows — ngxson · 2026-10-24
- Claude SSHes Into a Machine to Blink the Right Port When Asked Where to Plug In — peterjliu · 2026-10-24
- Security pros call AI security discourse incoherent: basics still win — AlexTensor · 2026-10-24