LLM chatbots reduce interaction effort for industrial decisions

A new mixed-methods study from Figliè et al. (arXiv:2605.31224) pitted LLM-based conversational interfaces against traditional graphical dashboards in industrial decision-making scenarios. Twenty participants used both interfaces to complete four simulated tasks of varying complexity. Researchers measured mental workload, completion time, and decision accuracy, supplemented by post-study questionnaires and semi-structured interviews analyzed via thematic analysis. The results show that conversational agents significantly reduced interactional effort by allowing users to query data directly through natural language, bypassing the learning curve of GUI navigation.

However, the graphical dashboard held its own for tasks requiring an overview, cross-referencing, or verification. The conversational interface excelled when users needed specific data points quickly, but lacked the spatial context a dashboard provides. The authors caution that these benefits may vary across tasks and that larger-scale validation is needed. For professionals, the takeaway is clear: LLM-based chat interfaces can complement, not replace, dashboards—optimizing speed for direct queries while keeping visual dashboards for holistic situational awareness.