Teaching Novice Computing and Programming in the Agentic AI Era [pdf] (cs.brown.edu)

Brown University researchers propose a transformative approach to computing education in the age of AI-driven programming agents. In their paper, titled "Teaching Novice Computing and Programming in the Agentic AI Era," they argue that AI programming tools—referred to metaphorically as “flaky compilers”—should not be seen as threats but as opportunities to enhance the curriculum. The authors advocate for integrating software engineering principles into introductory computing courses, emphasizing that students must learn to take responsibility for their AI-assisted creations. This shift is essential for equipping novice programmers with the skills to address the challenges posed by ambiguous natural language inputs and the unpredictability of AI outputs. The significance of this proposal lies in its call to rethink teaching methodologies for a new generation of programmers. By placing a greater emphasis on concepts like program specification, design plans, testing, and peer review rather than solely coding, the curriculum can better prepare students for real-world software development challenges. This approach acknowledges the complexities introduced by AI systems and aims to build a more robust understanding of quality assurance and usability from the outset. Ultimately, this shift could redefine educational standards in computing, transforming novices into responsible and innovative software engineers.