Rethinking the Methodology of Pair Programming in the Context of Intelligent Agent Integration
Sergei Kuznetsov · Global Society of Scientific Research and Researchers - International Journal of Computer · 2026
This article examines the transformation of pair programming methodology in the context of integrating intelligent assistants based on large language models. The study is conducted as a structured narrative review and analytical synthesis of academic publications devoted to AI-assisted programming, human–AI collaboration, and collaborative software development. The main focus is on changes in the structure of developer interaction, the redistribution of roles, and the reconfiguration of cognitive functions between human participants and intelligent agents. Key characteristics of traditional and AI-assisted pair programming are compared, including differences in evaluation parameters, learning effects, and behavioural outcomes. It is shown that the influence of intelligent assistants is not limited to improving productivity, but is realised through changes in the organisation and perception of the development process. It is also demonstrated that isolated use of AI tools does not fundamentally alter collaborative practices unless they are integrated into a coherent interaction structure. An original model of AI-integrated pair programming is proposed, describing the distributed nature of solution generation, evaluation, and refinement within a human–AI system. The results make it possible to consider pair programming as a hybrid form of collaboration, where the effectiveness of development depends on the coordination between human and intelligent components. The article may be of interest to software engineering researchers, educators in computing disciplines, and developers of AI-supported programming tools.