從助理到協作者: 生成式 AI 在人機協作智慧決策中的角色

Abstract

本研究探討生成式人工智慧（Generative AI）於組織決策中扮演不同社會角色所帶來的團隊互動效果。透過2×2因素實驗設計，操作AI的互動模式（個人助理 vs. 團隊協作者）與批判來源（人類反思 vs. AI回饋），共61位來自不同背景的參與者進行模擬決策任務。結果顯示，當AI被定位為「團隊協作者」時，顯著提升團隊參與度、討論效率與AI信任感，尤其在人類反思作為批判機制時，效果最為明顯。基於此，本研究提出「AI社會嵌入理論（AISET）」，指出AI若被視為具互動性、情境適應力與社會連結感的參與者，將有助於激發團隊動能並優化協作流程。研究對教育與企業實務皆具啟發性，建議未來AI應設計為具備社會互動能力的團隊成員角色，以提升組織內部決策的效率與創造力。

This study examines how the social role framing of generative AI influences team dynamics in organizational decision-making. Using a 2×2 factorial experimental design, 61 participants engaged in simulated strategy discussions under varying conditions of AI interaction (Personal Assistant vs. Team Collaborator) and critique source (Human Reflection vs. GPT Feedback). Results show that when GPT is framed as a Team Collaborator, participants report significantly higher levels of team involvement, discussion efficiency, and trust in AI—particularly when critique is provided through human reflection. To explain these effects, the study proposes the AI Social Embeddedness Theory (AISET), which conceptualizes AI as a socially embedded actor characterized by interactivity, contextual adaptability, and relational integration. Findings suggest that AI framed as an active team member can enhance cognitive engagement and collaborative performance. The research offers practical insights for both educational and organizational settings, encouraging the design of AI systems that support socially aware, team-oriented interaction patterns to improve decision-making quality and innovation potential.