An Adaptive AI Debugging Assistant for Computing Education
TL;DR
A context-aware Visual Studio Code extension that delivers progressively revealed, adaptive hints tailored to the user’s self-described experience level, indicating that carefully designed scaffolding successfully encourages active reflection and preserves problem-solving skills in computing education.
Abstract
Artificial Intelligence (AI) debugging assistants are increasingly prevalent in software development but present a pedagogical risk: "metacognitive laziness," wherein learners over-rely on automated solutions without grasping the underlying logic. To address this, we developed a context-aware Visual Studio Code extension powered by Gemini 1.5 Pro. Rather than providing direct answers, the tool delivers progressively revealed, adaptive hints tailored to the user’s self-described experience level. A user study yielded perfect bug resolution rates, and participants reported finding the tool useful, indicating that carefully designed scaffolding successfully encourages active reflection and preserves problem-solving skills in computing education.