Three instructor practices for introducing AI agents in SDPBL: foundation-first tutorials, continuous monitoring with timely intervention, and accountability for AI-generated code¶

Strategy · All strategies
Evidence · 1 claim (1 for) · 1 study (1 design), q2 · 0 of 1 report an effect size · 1 claim rests on one study

Description¶

Based on their SDPBL observations, the authors recommend three approaches for instructors adopting AI agents. The foundation-first approach designates early sessions as tutorials covering fundamentals, with reference sample applications so students can "understandin g and implement- ing code independently" before extensive AI use. Continuous monitoring uses regular assessments to detect comprehension drops early, since proactive intervention proved effective in lec09. Accountability for AI-generated code uses one-on-one interviews where students explain the logic behind any code they did not manually write, helping ensure implementation speed does not compromise understanding.

Design Implications¶

Context¶

Requirements¶

• Weekly one-on-one instructor interviews with each student to assess code comprehension on a defined scale

Constraints¶

• The authors note a significant negative correlation between AI usage and code comprehension was not observed in their data, so these practices are grounded in descriptive observations and instructor judgment

Target Learners¶

• third-year undergraduate students in team-based software development courses

Target Learning Goals¶

• maintaining code comprehension while using AI agents
• preventing over-reliance on AI-generated code

Claims¶

• Team code comprehension dropped in lec09 when Implementation-phase AI usage peaked, and recovered in lec10 following instructor intervention [+W]

Related Strategies¶

• Computing instructors should approach GenAI slide generation as programmers: work iteratively with coding assistants and text-based slide toolchains

Examples¶

-

Key Sources¶

• Hidetake Tanaka, Hiroshi Igaki, Kazumasa Shimari, Kiyoshi Honda, and Naoki Fukuyasu. (2026). Practical Implementation Report on Introducing Spec-Driven Development Using AI Agents in Software Development PBL. https://arxiv.org/abs/2608.30572