Collaborative Research: An Intelligent Coding Agent to Support Engaged Learning in Undergraduate Software Security Education

This project aims to serve the national interest by improving undergraduate software security education through the development and study of trustworthy artificial intelligence (AI) learning technologies that support engaged learning in secure coding. As software systems are at the backbone of critical infrastructure, preparing students to engineer secure software systems is an important national workforce priority. At the same time, AI assistants powered by large language models (LLMs), such as ChatGPT, are increasingly becoming part of undergraduate computing education, creating opportunities to provide personalized learning support while also raising concerns about inaccurate or unsafe guidance. This Level 1 Engaged Student Learning project plans to develop ERICA (Educational SecuRity-Focused Intelligent Coding Agent), an intelligent coding agent that helps students learn secure software development through formative feedback, adaptive scaffolding, and evidence-based instructional support. The project's significance lies in advancing understanding of how AI can be designed and responsibly integrated into undergraduate cybersecurity education while promoting student learning, instructor support, and broader adoption of effective educational practices. The project also plans to release ERICA, instructional materials, and supporting resources as open educational infrastructure to benefit educators across diverse institutional settings. This project’s goals are to investigate how trustworthy AI coding agents can support undergraduate secure coding education and to generate new knowledge about the design and classroom integration of AI-assisted learning technologies. The scope includes three complementary research thrusts. First, the project plans to investigate the knowledge boundaries of state-of-the-art large language models through a novel concept inventory aligned with undergraduate software security curricula, identifying the topics and reasoning tasks that current models perform reliably and where instructional support is needed. Second, the project plans to develop and evaluate a novel guardrail framework that promotes safe, resilient, and pedagogically appropriate classroom use of AI coding assistants. Third, the project plans to design, deploy, and iteratively refine ERICA, a human-centered intelligent coding agent that provides formative feedback and scaffolding during authentic secure coding activities. The project plans to study Erica's effectiveness through classroom deployments, qualitative and mixed-methods educational research, instructor and student feedback, and external evaluation to advance understanding of how AI can best support secure coding instruction. Open-source software, instructional materials, and dissemination activities across diverse institutional settings will facilitate broader adoption and future research in undergraduate software security education. The NSF IUSE: EDU Program supports research and development projects to improve the effectiveness of STEM education for all students. Through the Engaged Student Learning track, the program supports efforts to transform and improve STEM education across institutions of higher education and disciplinary communities. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.