Illusions of reflection: open-ended task reveals systematic failures in Large Language Models' reflective reasoning

Open in new window