Programming instructors remain largely unprepared for the reality of generative AI in their classrooms, despite ChatGPT's three-year public existence. Researchers conducting interviews with computing educators discovered that most instructors lack coherent strategies for adapting their pedagogy to a world where students routinely use AI coding assistants.
The problem runs deeper than surface-level resistance. Instructors face a genuine pedagogical crisis. Traditional programming curricula centered on having students write code from scratch no longer reflects how professional developers work. Yet shifting entire courses requires rethinking everything from assignment design to assessment methods.
Some educators have started experimenting with new approaches. A few institutions now focus on teaching students to effectively prompt and debug AI-generated code rather than write it from scratch. Others emphasize algorithmic thinking and problem decomposition, skills that remain difficult for AI systems to handle. Some instructors are redesigning projects to require integration of multiple systems, code review, and architectural decisions that demand human judgment.
But these responses remain scattered and ad hoc. No consensus exists on best practices. Departments lack institutional support for the time-intensive work of curriculum redesign. Training programs for new instructors still teach methods designed for a pre-AI era.
The stakes matter beyond the classroom. Graduates entering the workforce with weaker fundamental skills in code generation could struggle to contribute meaningfully in roles that assume those capabilities. Conversely, instructors who ignore AI entirely produce graduates unprepared for actual development environments.
What emerges is a pattern researchers call "emergency pedagogical design." Instructors improvise solutions to immediate problems rather than implementing thoughtful, comprehensive changes. Some ban AI tools outright. Others allow unlimited use with vague guidelines. Few have the bandwidth to develop middle-ground approaches that teach both AI collaboration skills and foundational programming knowledge.
The challenge persists because institutions have not invested in solving it. Universities need dedicated time,