Research suggests large language model outputs are still too opaque, biased and inconsistent for recommending scientific experts
Source: Computer Weekly