Human-in-the-Loop AI: A Case Study of Taiwan’s Digital Transformation

工商管理學系暨商學研究所

撰文者／Anoop Remanan Syamala

The fifth and final session of Fall 2025’s Operations Management Seminar I, hosted by Prof. Chia-Wei Guo, occurred on December 5th. The session began with a presentation by Assistant Professor Yu-Ling Hsu from the Department of Industrial Management at National Taiwan University of Science and Technology, Taiwan. Dr. Hsu shared a rich and experience-driven case on AI × Human Intelligence (AI × HI) collaboration, using Taiwan’s digital transformation journey as a concrete and longitudinal example.

Dr. Hsu began by situating the case historically, emphasizing that much of this work had started well before the current generative AI boom. At a time when AI was still largely rule-based and narrowly applied, her team was already thinking seriously about how intelligent systems could augment rather than replace human capabilities. Drawing on forecasts from organizations such as the World Economic Forum and McKinsey, she highlighted how early signals of workforce transformation motivated proactive experimentation in human–machine collaboration, rather than reactive automation following disruption.

A central theme of the talk was the distinction between automation-centric and human-centric digital transformation. Automation-centric approaches focus on replacing repetitive, hazardous, or highly structured tasks with machines, while human-centric approaches aim to enhance human perception, decision-making, and physical capability. Dr. Hsu illustrated this with vivid examples ranging from exoskeleton-assisted labor and AR-supported factory operations to precision surgery and human-sensing technologies. These cases made it clear that productivity gains do not have to come at the expense of human agency; instead, well-designed systems can shift human effort toward higher-value tasks.

The conceptual backbone of the presentation was the idea of AI × HI collaboration, later operationalized through digital twin architectures. Dr. Hsu explained that digital twins are not merely visual simulations, but dynamic systems that integrate real-time sensor data, historical records, predictive models, and decision support. Notably, her team proposed an expanded framework that goes beyond physical-device twins to include human twins, process twins, and system-level twins, explicitly embedding human behavior and judgment into the digital representation. This multi-layered architecture reframed digital twins as coordination mechanisms across people, machines, and organizations, rather than purely technical artifacts.

The most compelling part of the seminar was the in-depth case study on Taiwan’s aquaculture industry. Dr. Hsu walked the audience through how AIoT sensors, real-time water-quality monitoring, image recognition, and cloud-based analytics were combined to digitalize tacit knowledge held by experienced fishery workers. The system not only improved survival rates and production stability but also addressed deeper structural challenges such as aging labor, knowledge loss, and the difficulty of training new entrants. The discussion demonstrated how digital transformation can serve as a vehicle for preserving and transferring knowledge, as well as achieving efficiency gains.

Throughout the Q&A session, Dr. Hsu repeatedly emphasized the importance of human-in-the-loop design. While AI systems can optimize, predict, and recommend, she argued that humans must remain responsible for interpretation, correction, and ethical judgment—especially given the black-box nature of many AI models. This stance resonated strongly with operations management concerns about accountability, explainability, and decision ownership. Her reflections on applications in biotechnology, pharmaceuticals, manufacturing, and even sports further reinforced that AI × HI collaboration is most valuable in high-risk, high-value contexts where the cost of failure is substantial.

In conclusion, the seminar served as a fitting culmination of the course. Rather than presenting AI as a disruptive force to be managed defensively, Dr. Hsu offered a forward-looking and pragmatic vision in which AI becomes an enabling infrastructure for better human decisions. The session tied together themes from technology, operations, and human factors, leaving a strong impression that successful digital transformation is not about choosing between humans and machines, but about designing systems where they learn, adapt, and perform better together.