Grading AI answers: build an LLM evaluation rubric

Rate and correct 200 chatbot answers for accuracy, tone and clarity; write the rubric and a findings note.