Nathanpeck (auto-discovered)

A test by an unnamed author shows that large language models like Claude Fable tailor their responses to the user's prompt style, often resulting in condescending output. When asked to communicate wit…