Innovation in Human-Generative AI Collaboration for Software Development

Abstract

This paper explores human-generative AI collaboration, with a focus on tacit knowledge and mutual knowledge transfer via the novel approach of AI pair programming in software development. Academically, the paper identifies the emergence of a new organizational capability, which can be termed ‘generative AI-based IT capability,’ and examines its manifestation mechanism. At its core lies a unique bidirectional knowledge transfer process between humans and AI. Specifically, this consists of: the ability to externalize knowledge including the tacit dimension in an AI-interpretable manner, and recognition of the limitations of this process; as well as the ability for humans to actively and contextually interpret and evaluate the rationale of AI-generated explicit knowledge (such as code), and to internalize it into their own understanding. Practically, the paper presents collaborative methods, guidelines for AI technology-driven software development, and insights for human resource education.