One powerful way to do this is through a routine called slow reveal graphs.

Research suggests Large Language Model outputs are still too opaque, biased and inconsistent for recommending scientific experts ...