Summary
Medical question answering systems that work across different languages might give different answers depending on cultural context or stay consistent everywhere. The authors looked at many experts from medicine, language technology, and anthropology across several countries to see if answers should be universally the same or adjusted culturally. Anthropologists preferred adapting answers to local cultures, while doctors and language experts were split, with differences especially between the US and Europe. Current language models tend to ignore these cultural differences and favor consistent answers. The authors found that it is unclear which approach best serves users and more research is needed.
What this means in practice
- •For medical ai developers: Design multilingual medical language models that consider cultural differences when answering clinical questions.
- •For health policy makers: Inform guidelines on the use of AI for medical advice across different cultural and linguistic populations.
A survey. It maps existing work.
Authors
Minh Duc Bui, Mario Sanz-Guerrero, Abteen Ebrahimi, Sagi Shaier, Peter Herbert Kann, Manuel Mager, Katharina von der Wense
Abstract
Should multilingual LLMs answer medical questions consistently across input languages, or adapt responses to cultural cues? Existing multilingual medical benchmarks usually assume that medically correct answers should remain consistent across languages and treat cross-lingual variation as model error. In contrast, cultural adaptation research argues that appropriate medical answers may legitimately differ across contexts. We review the multilingual medical NLP literature through these two perspectives, we identify three gaps: limited stakeholder perspectives (e.g., of medical professionals), a lack of empirical evidence on which approach better serves users, and no benchmarks capable of distinguishing universally correct from culture-specific cases. To address the first gap, we survey 356 participants across three stakeholder groups (medical, NLP, and anthropology professionals) in three countries (Germany, Spain, and the United States). Anthropologists consistently favor adaptation, while medical and NLP respondents remain divided, with notable divergence between U.S. and European medical professionals. LLMs prompted with profession and country personas fail to reproduce this variation, overestimating cross-lingual consistency preference among NLP and medical personas. We conclude that neither consistency nor adaptation can currently be considered clearly preferable, highlighting the need for empirical evidence on which approach better serves users across cultural contexts.