AI LLMs Astonishingly Bad At Doing Proofs And Disturbingly Using Blarney In Their AnswersApril 8, 2025