Generative AI tools are rapidly transforming software development practice, prompting unprecedented research interest. However, existing studies have predominantly examined initial adoption rather than sustained use. Understanding what drives developers to continue using these tools after initial adoption remains underexplored, particularly in small and medium-sized enterprises where resource constraints shape technology decisions differently than in large organisations. This study investigates factors associated with developers' intentions to continue using GenAI tools, adapting the UTAUT2 framework to post-adoption professional contexts. We employed a two-phase mixed-methods design. Phase 1 comprised a six-month longitudinal pilot study at an Italian software company combining surveys and interviews with 17 developers to explore how perceptions of GenAI evolve as experience accumulates. These insights informed a structural model tested in Phase 2 through a cross-sectional survey of 154 developers across Italian SMEs, analysed using PLS-SEM. The model explained substantial variance in continued use intention (R2 = 0.647), with individual-level perceptions, particularly around productivity, enjoyment, and ease of use, driving sustained adoption, whereas social and organisational factors played no significant role. These findings suggest that, for GenAI tools, post-adoption behaviour differs from initial adoption patterns: in voluntary professional contexts, sustained use is driven primarily by individual-level factors rather than by social and organisational support.

翻译：生成式人工智能工具正在迅速改变软件开发实践，引发了前所未有的研究兴趣。然而，现有研究主要关注初始采纳而非持续使用。在初始采纳后，驱使开发者继续使用这些工具的因素仍未得到充分探索，尤其是在资源约束对技术决策的影响与大型组织不同中小企业中。本研究调查了与开发者继续使用GenAI工具意图相关的因素，将UTAUT2框架调整至采纳后的专业情境。我们采用了两阶段混合方法设计。第一阶段包括在意大利一家软件公司进行的为期六个月的纵向试点研究，结合调查和访谈，涉及17名开发者，以探索随着经验积累，对GenAI的看法如何演变。这些见解为第二阶段测试的结构模型提供了依据，该模型通过对意大利中小企业154名开发者进行的横截面调查进行测试，并使用PLS-SEM进行分析。该模型解释了继续使用意图的显著方差（R² = 0.647），其中个体层面的感知（特别是围绕生产力、享受和易用性）驱动了持续采纳，而社会和组织因素未发挥显著作用。这些发现表明，对于GenAI工具，采纳后的行为与初始采纳模式不同：在自愿的专业情境中，持续使用主要由个体层面因素驱动，而非社会和组织支持。