Should Cursor be taught in college? (ben.page)

A recent debate sparked by Andrew Mayne’s conversation on the OpenAI Podcast asks whether tools like Cursor—AI-driven, agentic code assistants—should be taught in college. The author, a recent graduate whose CS program focused on C and C++, argues no: colleges should prioritize long-lasting fundamentals over transient tool-specific training. They note that practical frameworks (React) and even recent breakthroughs (ChatGPT, Cursor, model versions like “gpt-5-codex” or “sonnet-4.5”) can appear or change within a student’s degree, making curricula that teach prompting tricks or current agentic workflows obsolete quickly. For the AI/ML and education communities this is a critical tension: how to prepare students for immediate industry tools without sacrificing enduring concepts. The practical takeaway is to teach core CS principles and mental models that make future tool adoption easier, while offering optional short-format learning (clubs, workshops, or elective labs) for hands-on agentic coding. That approach preserves academic longevity and builds adaptability—arguably the most valuable skill as models and interfaces evolve at a breakneck pace.