CS EDUCATORS OVERHAUL EXAMS AS AI CODING TUTORS RESHAPE TESTING

The rise of AI coding assistants has forced a fundamental reckoning in computer science education. Educators are pivoting from testing students' ability to write code to assessing their understanding of programming concepts and problem-solving approaches. Oral exams and proctored testing have emerged as primary strategies to verify student knowledge in real-time. Project-based work offers another avenue, allowing educators to evaluate skills through extended, supervised assignments. Despite widespread changes, the transition lacks guidance. Nearly half of surveyed educators report they have no proven examples for effectively integrating AI into their courses, creating a vacuum in best practices. The paradox centers on a core tension: AI tutors can now generate correct code instantly, making traditional assessments obsolete. Yet educators struggle to establish which skills truly matter when code generation is commoditized. The profession faces a critical need for shared frameworks and curriculum models that account for this new reality.