Calibrated? Not for Everyone: How Sexual Orientation and Religious Markers Distort LLM Accuracy and Confidence in Medical QA
Alberto Testoni, Iacer Calixto · 2026
Read Calibrated? Not for Everyone: How Sexual Orientation and Religious Markers Distort LLM Accuracy and Confidence in Medical QA by Alberto Testoni, Iacer Calixto (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-17316.txt.
Description
Testoni and Calixto test whether patient identity cues warp medical question-answering by language models. They ran nine general and medical models over 2,364 questions plus altered versions. Markers for homosexual patients repeatedly lowered scores, and combined identities caused uneven damage to accuracy and confidence. A clinician-checked open-ended case confirmed the pattern was not just a multiple-choice artifact.
Catalog record
| CREATOR | Alberto Testoni, Iacer Calixto |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 57802 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.17316