Applying IRT to Distinguish Between Human and Generative AI Responses to Multiple-Choice Assessments