Mindsets, cultures, and technologies: supporting end-user development in the AI era

The rapid diffusion of AI exemplified by Large Language Models (LLMs) has intensified long-standing tensions between automation-centric approaches and visions of human-centred empowerment. This paper argues that End-User Development (EUD) is no longer an optional enhancement, but a socio-technical necessity for coping with wicked problems. To address this challenge, the paper integrates four complementary socio-technical frameworks into a unified conceptual foundation for EUD in the AI era. Building on this synthesis, we introduce the Mindsets-Cultures-Technologies (MCT) model, which explains EUD outcomes as a function of alignment across these three dimensions. This model provides a principled account of why identical AI systems can either empower users or reinforce dependency. Using this integrative lens, the paper critically analyses both the opportunities and risks posed by LLMs for EUD. While LLMs can lower barriers to participation and support learning-on-demand, their unreflective use risks reinforcing consumer mindsets, reducing productive struggle, obscuring design trade-offs, and fostering overreliance on technology. We argue that these risks stem less from LLM capabilities than from how they are embedded in socio-technical environments. The paper concludes with a set of design principles for AI-augmented EUD, positioning EUD and the MCT model as foundational requirements for coping with wicked problems.