2026 ASEE Annual Conference & Exposition

Introduction: In pair programming, two people work together on a programming task, with one serving as the “driver” and the other as the “navigator.” Research on pair programming has been somewhat mixed: in many cases it is an effective approach for promoting positive learning outcomes, but it may not be beneficial to all students. Generative AI has created a new possibility for pair programming: pairs composed of one student and an AI tool. Despite its novelty, there is a growing body of research on AI pair programming.

Objective: The objective of this study is to answer the research question: What evidence exists regarding the outcomes of student-AI pair programming?

Methods: We queried five traditional collections of academic research; we also prompted two AI tools designed to identify academic research. We identified 10 relevant studies. We analyzed each to determine the context (e.g., grade level), the AI tool(s) used, the research type and design, what (if any) preparation was provided to participants, what outcomes were considered, and the study’s findings.

Results: Almost all studies involved college students. The studies used a variety of methods – from analysis of interview transcripts to surveys for affective constructs. The most commonly used AI tool was ChatGPT or a variant of it. Study findings varied, but there was evidence of positive learning outcomes paired with less social connection when AI tools were used. Other affective outcomes – such as engagement – could be enhanced or reduced when students were paired with an AI tool. Some findings indicated a preference for a human partner.

Discussion: More research is needed to support these findings and to extend them to other contexts. Specifically, it is important to conduct research that explores how the factors important in traditional pair programming impact AI pair programming. Capturing the cognitive advantages that AI tools might offer students while mitigating the potential social harms will be an important part of the responsible development of AI pair programming as a pedagogical approach. We also comment on the performance of the various databases and AI tools used to locate relevant research.

AI Use Disclosure: As noted above, AI was used to locate articles for the systematic literature review.