Teacher argues AI tutors should preserve student cognitive effort

Scott Farrar, a math teacher, discusses the intersection of AI and math education, emphasizing the importance of maintaining student agency and cognitive effort. He contrasts his household AI setup, where he retains human-in-the-loop control, with AI tutoring tools like Muse and Instinct. Farrar argues that effective AI tutors should not simply provide answers or short-circuit the learning process but should instead extract knowledge from students, encouraging them to explain their reasoning.

He describes a model where the AI acts as a 'polite demander' or 'warm demander,' prompting students to articulate their own understanding rather than passively receiving information. This approach leverages AI's infinite patience to help students navigate the 'zone of frustration' or 'zone of proximal development' without removing the productive struggle necessary for learning. Farrar notes that students often feel safer sharing their thought processes with AI than with humans, making this interactive, agency-preserving model particularly valuable for education.