i wonder if one would find some clues/pointers towards aligning LLMs by studying the way people's valence/CDNS profile influences the language they use

Posters: cube_flipper. cube_flipper

2 likes recorded by the source corpus.