I feel like debates about AI and plagiarism are quite surface-level definitional debates. They miss out an important discussion of why we believe (old-fashioned) plagiarism is bad. Answers about when using AI is acceptable should be based on that reasoning.