Collaborative research: CER: enhancing novice programmers' AI help-seeking through metacognitive interventions

The National Science Foundation Division of Computing and Communication Foundations awarded CAL Poly Corporation $170,791 on October 1, 2026, under the Computer and Information Science and Engineering program to develop instructional materials and software tools that teach novice programmers how to use artificial intelligence systems effectively for learning.

The project addresses how beginning computer science students often misuse AI tools like ChatGPT and GitHub Copilot by requesting complete solutions or copying code without understanding it. The research team will develop classroom curriculum teaching metacognition—awareness of one's problem-solving process—to help students construct more pedagogically sound AI help requests appropriate to their current problem-solving stage. The team will also prototype a web extension that tracks AI usage and provides real-time feedback to discourage unproductive help-seeking and encourage metacognitive awareness. These outputs equip computer science teachers and students with resources to foster AI use that supports rather than impairs learning, addressing workforce readiness for problem-solving in AI-augmented environments.

Work is performed in San Luis Obispo, California, with a period of performance from October 1, 2026, through September 30, 2029. CAL Poly Corporation, a non-profit auxiliary organization supporting California Polytechnic State University, receives the award as the parent recipient.

Abstract

Students learning computer programming today often turn to artificial intelligence (AI) systems like chatgpt and github copilot for help. Unfortunately, they often use AI in ways that reduce learning, such as asking for full solutions, copying code they do not understand, or even asking AI to solve the wrong problem. This project aims to teach beginning programmers how to ask AI for help in ways that support learning rather than impair IT. The key idea is to improve students' awareness of their problem-solving process and to teach them how to use this knowledge to construct more effective AI help requests. This project will produce new evidence-supported instructional materials and software tools for computer science teachers and students. IT addresses the national need to prepare students for a workforce where strong problem-solving skills are essential for working effectively with AI and evaluating the quality of its outputs. In order to foster the kind of AI use that promotes the learning of fundamental computer programming practices, this project adopts novel interventions targeting metacognition. Metacognition is the awareness of one's problem-solving process, including the steps needed to solve a problem, the current step, the strategies adopted for each step, and the outcomes of using such strategies. The research team will develop a classroom curriculum that teaches metacognition for solving programming problems and encourages students to construct AI help requests that are both pedagogically sound and appropriate to their current problem-solving stage. To further enable ongoing practice and achieve sustained behavioral change, the researchers will prototype a web extension that tracks AI usage and provides real-time feedback. The feedback will discourage help-seeking that simply asks for the answer, encourage the use of metacognitive knowledge, and provide tailored suggestions for how to ask for help. All interventions will be evaluated in the lab and authentic classroom settings for how they change students' behaviors, metacognitive awareness of problem-solving, and learning outcomes in the ability to independently solve programming problems. 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.- subawards are not planned for this award.