Avoiding the Automation Trap: Designing Human-Centred AI Workflows for Dental Manufacturing

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This industrial case study explores integrating machine learning into high-volume dental manufacturing. Such technologies profoundly influence existing technician workflows and laboratory operations. To address this, a legacy software audit using a localised Cognitive Dimensions of Notation (CDN) framework identified high Repetitive Viscosity, Hidden Task Status, and Workflow Rigidity. This diagnostic phase was driven by structured workshops with the digital technicians. These failings hindered quality control and forced technicians into redundant tasks. To resolve these issues, the project co-designed a human-in-the-loop application that replaced modal menus with a unified 3D editor layout, eliminated hidden states with a visual column-based progress queue, and introduced state serialisation to support flexible task management. Qualitative evaluation confirmed that these design choices effectively resolved the identified cognitive frictions by reducing interaction costs and preventing repetitive editing tasks. While improvements to backend algorithms halved baseline processing time, successful digital transformation relies equally on workflow redesign and end-user engagement. Ultimately, this method of adapting, localising, and running workshops provides a technique that can be deployed across a wide range of industrial settings when faced with the challenge of the AI-assisted automation trap. This is of particular importance in industrial sectors where machine learning-based solutions risk threatening existing work practices and disrupting established processes.