FROM CODING TO AI-ASSISTED DEVELOPMENT: EXAMINING DEVELOPERS’ ACCEPTANCE OF GENERATIVE AI THROUGH AN EXTENDED TAM

Abstract

The swift adoption of AI in software development is reshaping traditional coding methodologies and presenting novel opportunities for AI-driven development. But besides the technical qualities of the technology, developers need to consider perceptions of the usefulness, ease of use, trust and risk. In this study, an extended Technology Acceptance Model (TAM) was used to explore the acceptance of generative AI-based development by developers. The model included perceived usefulness, perceived ease of use, perceived risk, and perceived trust as antecedents of the attitude, and behavioral intention and use of the AI-assisted development followed. The quantitative cross-sectional design was used, and a purposive sampling technique was used to collect data from software developers, software engineers, programmers and IT professionals of Karachi and Hyderabad, Sindh, Pakistan, a total of 229 respondents were sampled. Partial Least Squares Structural Equation Modeling – SmartPLS was used to analyze the data. The results showed that perceived usefulness, perceived ease of use, and trust had a positive effect on developers' attitudes, while perceived risk had a significant negative effect. The attitude was found to have a strong influence on the behavioral intention, which, in turn, predicted the use of AI-assisted development. The mediation effect and sequential mediation of attitude and behavioral intention was also confirmed, further explaining the mediation results obtained. The results of the mediation were further confirmed by the significant mediation and sequential mediation of attitude and behavioral intention. The study reveals that developer-centric strategies for sustainable software development are required, as developers' perceptions of both facilitating and perceived barriers to the adoption of generative AI affects their adoption.