Leading AI models ace many vaccine questions but falter on clinical rules
A multilingual benchmark of 1,886 vaccine-related questions found that large language models answered most items accurately across English, Spanish, and Chinese. However, performance varied by vaccine type, language, and prompting…