知识融合视角下人机协同效用测度研究OACHSSCD

[Purpose/Significance]In today's digital and intelligent era,human-AI collaboration has become not just a technical idea but a national strategic priority.Yet most current research still focuses on technology and efficiency,over-looking the deeper issues of how knowledge is integrated,how humans and AI build shared understanding,and how emo-tions shape their interactions.To fill this gap,this study develops a way to measure the real value of human-AI collabora-tion from a knowledge integration perspective,aiming to uncover what actually happens in the process black box and address common problems like knowledge flow breakdowns and poor adaptation to changing situations.[Method/Process]We built a three-dimensional model based on the Cognitive-Affective-Behavioral framework to break down the process of knowledge integration in human-AI collaboration.We then turned this model into an evaluation system with 18 first-level and 44 second-level indicators,covering cognitive interaction,emotional adaptation,and behavioral coordination between humans and AI.To assess collaboration utility,we used an improved AHPSort Ⅱ method.We introduced interval-valued intuitionistic fuzzy sets to handle uncertainty in expert judgments and added a conflict resolution factor to help experts with different backgrounds reach agreement.We also incorporated a logistic function to better capture the nonlinear nature of collaboration,especially when factors like trust build up gradually.We tested the framework on a high-tech manufacturing company.We collected data from company reports,expert evaluations through a Delphi survey,and industry benchmarks.We calculated indicator weights,processed mixed data with the enhanced AHPSort Ⅱ method,and determined the overall collaboration level.[Result/Conclusion]The results place the company's human-AI collaboration at a highly synergistic level.But a closer look reveals an imbalance:efficiency and innovation score high,while synergy and risk control score much lower.This points to a collaborative void-interactions between humans and AI stay superficial,lacking deep know-ledge fusion or real cognitive alignment.Our framework proves useful in identifying such specific bottlenecks.The study makes three main contributions.First,it offers a holistic way to understand human-AI collaboration through the lens of knowledge integration,moving beyond single-dimension views.Second,it provides a practical method that combines an enhanced AHPSort Ⅱ with fuzzy set techniques to handle mixed data and dynamic classification.Third,it gives organiza-tions actionable insights to move human-AI interaction beyond simple tool use toward truly intelligent and synergistic team-work.Future work could focus on making the model more adaptive through self-learning algorithms and testing it across diffe-rent industries and cultural settings.