智能组织中的人机协同决策：基于人机内部兼容性的研究探索

智能时代已然来临。智能技术的发展正加速推动组织的智能化进程。越来越多的企业在生产和管理中采用智能技术以提高竞争力。在此背景下, 人机协同工作日益普遍, 人机协同决策成为新型组织决策方式。然而, 当前智能组织中的人机协同决策还面临信任度低、可控性低、透明度低、协同度低等一系列问题, 它们阻碍了决策质量、效率和体验的提升。本项目认为, 人机兼容性, 特别是人机内部兼容性, 如认知兼容性、情感兼容性、价值兼容性等, 或是影响人机协同决策绩效的根本原因。因此, 本项目基于人机内部兼容性理论视角, 综合采用决策心理学、认知科学、组织行为学等多学科理论与方法, 通过一系列现场研究和模拟实验, 力图揭示人机协同决策中存在的问题及其成因, 探究人机协同决策中内部兼容性的影响因素和作用机制, 进而提出若干人机协同决策优化方法。项目研究成果将有助于促进人机协同决策理论与人机兼容性理论的发展, 提升人机协同决策绩效, 推进组织决策的智能化进程。

Human-agent collaborative decision-making in intelligent organizations: A perspective of human-agent inner compatibility

The era of artificial intelligence has already arrived. With the rapid development of intelligent technology, more and more companies are adopting this technology into their business processes to enhance their core competitiveness. Subsequently, human-agent collaborative work is becoming common, and human-agent collaborative decision-making (HACDM) is evolving as a new form of organizational decision-making. However, evidence shows that HACDM still faces challenges, such as low trust and controllability toward agents, low transparency of agents, and low collaboration between humans and agents. Therefore, how these challenges can be overcome to improve the decision quality, decision efficiency, and user experience of HACDM is crucial to the field of organizational decision-making.This project suggests that human-agent compatibility, especially human-agent inner compatibility (HAIC) which consists of cognitive, affective, and value compatibility, might be the fundamental factor affecting the performance of HACDM. Following the perspective of HAIC theory and using the multi-disciplinary methods from psychology, cognitive science, and organizational behavior, we intend to 1) reveal the existing problems within HACDM; 2) explore the impact of HAIC on the process and performance of HACDM; 3) propose methods to improve the performance of HACDM. Thus, this project consists of three studies. Study 1 aims to investigate real-world intelligent organizations to uncover the current usage of agents, the willingness of human employees and managers to collaborate with agents, and the possible problems within HACDM. Based on the findings of study 1, study 2 adopts HAIC theory as its framework and explores the influence of cognitive, affective, and value compatibilities on the process and performance of HACDM. Finally, study 3 tests the effectiveness of the several methods suggested by HAIC theory for improving HACDM, such as increasing the transparency of agents’ decisions and providing decision feedback to human employees.This project’s findings will contribute both theoretically and practically. Theoretically, this project examines the components of HAIC (i.e., cognitive, affective, and value compatibilities) and investigates their influence on HACDM. Thus, it will contribute to the further development of human-agent compatibility theory and human-agent collaboration theory. Practically, the project proposes several methods that can effectively improve the performance of HACDM. Therefore, it will improve the performance of intelligent organizations and promote the intelligentization progress of HACDM.