AI Where It Matters: Where, Why, and How Devs Want AI Support in Daily Work (arxiv.org)

Researchers report a large-scale, mixed-methods study of 860 developers that produces the first task-aware, empirically validated map of where, why, and how developers want AI support in daily software work. Using cognitive appraisal theory to link developers’ task evaluations to AI openness and use, the study identifies distinct patterns: strong current use and desire for better AI in core engineering work (coding, testing), high demand for automation to reduce toil (documentation, operations), and clear resistance to AI in identity- or relationship-centric tasks (mentoring, hiring). The paper also quantifies contextual responsible-AI priorities—reliability and security for systems-facing tasks; transparency, alignment, and steerability so developers keep control; and fairness and inclusiveness for human-facing work. For the AI/ML community and tool builders, the implications are concrete: design context-aware assistants that emphasize robustness and safety in infrastructure tools, provide explainability and fine-grained controls for code and design support, and avoid or radically reframe automation in interpersonal domains where human judgment matters. The task-level breakdown gives product teams empirical guidance on where investment will be most adopted and trusted, and signals that responsible-AI features must be prioritized differently depending on whether the tool touches systems, code, or people.