Cheng-Ta Yang
Department of Psychology, National Cheng Kung University
Abstract
The rapid advancement of artificial intelligence (AI) is transforming how humans interact with technology, making human–AI collaboration a prominent research topic. However, previous research has shown that working with AI does not necessarily improve task performance; the effectiveness of collaboration depends on how humans and AI integrate information and make decisions across different contexts. AI accuracy and confidence, task difficulty, and the type of information provided by AI can interactively influence the effectiveness of collaboration. This talk will introduce Systems Factorial Technology (SFT), a rigorous theory-driven methodology for characterizing decision processes by their underlying mental architecture, workload capacity, and decisional stopping rules, to investigate human–AI collaboration. Within this framework, we can identify how information from humans and AI is integrated, whether it is processed in parallel or sequentially, and how processing capacity changes when AI assistance is introduced. Our results shed light on the effects of AI accuracy under varying levels of task difficulty, different approaches for manipulating task difficulty, and the influence of metacognition—particularly AI confidence—on decision-making efficiency. Finally, we will present applications of this research series in medical decision-making and Deepfake detection, and provide practical guidelines for effective human-AI collaboration.
