Metraipolis › archive › dc_cfb392dcfd45

Demographic Injection in Medical Language Models under Diversity, Equity, and Inclusion Prompts — arXiv:2608.15254

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.15254
License (as archived): CC0-1.0
Archived: 2026-10-18T17:13:32.436Z
SHA-256 of the archived text: 7b1ca49be42eb9a48da05d786903b9003d0af3087fca24afcf9f73398c8673d1
Stable URL of this page: https://metraipolis.com/documento/dc_cfb392dcfd45

How to cite

Demographic Injection in Medical Language Models under Diversity, Equity, and Inclusion Prompts — arXiv:2608.15254 — arXiv (en), https://arxiv.org/abs/2608.15254 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cfb392dcfd45 (Metraipolis archive dc_cfb392dcfd45, sha256 7b1ca49be42eb9a48da05d786903b9003d0af3087fca24afcf9f73398c8673d1)

BibTeX

@misc{dccfb392dcfd45,
  title = {Demographic Injection in Medical Language Models under Diversity, Equity, and Inclusion Prompts — arXiv:2608.15254},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.15254},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cfb392dcfd45; sha256 7b1ca49be42eb9a48da05d786903b9003d0af3087fca24afcf9f73398c8673d1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15254. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Clinical-AI guidance increasingly recommends prompting language models to reason with attention to diversity, equity, and inclusion (DEI). We measure a side effect that misrepresents patients: a one-sentence DEI prompt appended to a medical question leads models to add patient demographic attributes (race, socioeconomic status, sex) the question never stated, in effect rewriting who the patient is. We call this demographic injection. Across 47 models, four medical benchmarks, and 376,000 responses scored by a validated model-judge pipeline, a single DEI prompt raises the injection rate from 0.7% to 33.1% (47x) in all 47 of 47 models, attributable to the equity content rather than to added length (18x above a length-matched control; p=1.4x10^-14). Most added content is a general population statement that leaves the answer unchanged, but a smaller subset attaches an attribute to the specific patient or changes the selected option (0.25-2.4% of responses, 99.8% toward the incorrect option), where the invented demographic changes the answer the model recommends. Phrasing scales the effect from 14% to 56%. DEI prompts are just one example of a more general mechanism. Any instruction that nudges how a model reasons can make it add unrequested details, including details about the patient. Flagged outputs are treated as model errors under study, not clinical guidance.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.15254 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-18T16:29:13.780Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).