Teaching with the Machine: Designing AI ‐Augmented Novices to Foster Critical Thinking and Computational Thinking in STEM Education

This chapter explores two Columbia University case studies that show how AI can deepen student learning. The first case, in chemical engineering, introduces a deliberately flawed chatbot designed to mimic a confident but error-prone student. By critiquing its answers, students practice one of the hardest skills to teach in engineering: critical thinking. The second case, in computer science, examines how generative AI in introductory programming courses changes learning objectives and pedagogical strategies. Together, these examples suggest a broader framework for AI-augmented learning: using AI to promote critique, synthesis, and higher-order reasoning rather than shortcuts. STEM education faces a particularly sharp dilemma. On the one hand, generative AI can instantly solve many traditional homework and exam problems, raising concerns about academic integrity and assessment validity. On the other hand, the very flaws of AI—its tendency to hallucinate, confuse definitions, or overstate answers—can be transformed into teaching opportunities.