Human-AI collaboration in semiconductor manufacturing: socio-technical capability priorities and perspectives across organizational roles in Korea

AI transformation in semiconductor manufacturing requires choices among competing capabilities, yet little is known about how those priorities differ across organizational roles. This study combines the STOPE (Strategy, Technology, Organization, People, Environment) framework with a sequential Delphi-AHP design and draws on the judgments of 29 senior practitioners from memory manufacturers, equipment and materials suppliers, and a technology consultancy in Korea. The analysis produced two main findings. First, the People dimension received the highest ratings in Delphi, whereas Technology received the highest AHP weight (0.264) once experts had to trade capabilities against one another. Within Technology, real-time manufacturing data infrastructure (global weight 0.0998) and AI-robotics convergence (0.0909) ranked first and second. Second, strategy planners assigned a higher weight to People than Manufacturing/Research experts did (U = 50.0, p = .021, r = .43). The difference does not survive correction for multiple comparisons and is reported as exploratory. The results are organised into a strategic priority framework for high-reliability fab environments. In the panel’s judgment, data infrastructure is the base on which other AI capabilities rest, and human final judgment, expressed as an ‘AI-as-Co-Pilot’ principle, remains a capability to be developed further.