Generative AI in the Developer's Toolkit : Adoption and Factors Influencing Continued Use Among Software Developers
Year: 2026
Type: dissertation
Abstract: Generative AI coding tools such as GitHub Copilot, ChatGPT and Cursor have rapidly become embedded in professional software development, yet little is known about how individual developers adopt these tools in practice and what determines whether their use is sustained over time. Existing research has largely focused on technical performance and productivity benchmarks, leaving the experience of adoption underexplored. This study investigates how software developers adopt generative AI coding tools and which factors shape their continued use in professional practice. The study takes a qualitative interpretive approach, drawing on five semi-structured interviews with practicing software developers of varying seniority and across different organizational contexts. The data were analyzed thematically using Braun and Clarke's six-phase method and interpreted through the Expectation-Confirmation Model (ECM), with trust included as a complementary construct. The findings show that adoption patterns vary considerably and are shaped by a combination of personal initiative and organizational policy. GitHub Copilot was used by all participants alongside conversational tools, primarily for boilerplate, debugging and learning their way around unfamiliar codebases while complex or security-sensitive work was largely kept under human control. Continued use was driven mainly by workflow fit, calibrated trust developed through routine output verification and the constraints imposed by organizational policy. A concern about long-term skill development emerged particularly among junior developers and is not captured by standard adoption models. The study contributes to the application of ECM in the context of AI coding tools by arguing that trust, organizational factors and long-term skill development should be incorporated alongside the model's core constructs.
Author: Budak, Berkant
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7165820148