UNDERSTANDING HUMAN-AI COLLABORATION –   FROM ANTHROPOMORPHIC PERCEPTIONS TO WORKPLACE EXPERIENCES AND INTERVENTION STRATEGIES

Yingwei (Wayne) Xu

This dissertation advances understanding of artificial intelligence (AI) in service contexts by examining how AI anthropomorphism and human–AI collaboration shape employee outcomes through a multi-study approach.The first study develops a holistic conceptualization of AI anthropomorphism and introduces the Scale of AI Anthropomorphism (SAIA). Building on Gestalt theory, this research conceptualizes anthropomorphism as a multidimensional construct capturing both external and internal human-like traits, while also acknowledging both human virtues and vices. Following a six-step scale development procedure, including conceptualization of the construct, specification of the measurement model, item generation, item refinement using two rounds of exploratory factor analysis (EFA), evaluation of the latent structure using confirmatory factor analysis (CFA), and assessment of nomological validity using structural equation modeling (SEM), data were collected across five studies with a total sample of 2,944 participants. A 40-item, six-dimensional scale is established, encompassing human-like appearance, cognitive competency, adaptive capacity, social intelligence, morality, and fallibility. The results show strong internal consistency, convergent and discriminant validity, as well as stable factor structures across different AI systems. In addition, nomological validity results indicate that different AI anthropomorphism dimensions have distinct effects on individuals’ usage intentions across different service contexts and AI systems. The second study examines how AI type, anthropomorphism, and task complexity jointly influence employee workplace well-being in human–AI collaboration. Grounded in the Job Demands–Resources model and the Service-Profit Chain model, this research consists of three experimental studies. Study 1 examines the main and interaction effects of AI type, anthropomorphism, and task complexity on employee workplace well-being. Study 2 investigates the mediating roles of identity threat, creativity, and morale. Study 3 examines whether employee workplace well-being influences customer outcomes. The data were analyzed using a series of ANCOVA tests and mediation analysis using Hayes’ PROCESS Model 4. Results show that employees experience the highest well-being when collaborating with low-anthropomorphic chatbots on complex tasks, whereas highly anthropomorphic robots performing routine tasks generate the lowest well-being. Furthermore, while identity threat, creativity, and morale play significant mediating roles, identity threat is primarily influenced by AI anthropomorphism and AI type, whereas creativity and morale are primarily driven by task complexity and AI type. In addition, higher employee well-being leads to more favorable customer outcomes. The third study investigates employees’ willingness to collaborate with AI by identifying key inhibitors and underlying mechanisms. Drawing on Technology Threat Avoidance Theory, this research consists of three studies. Study 1 uses focus group discussions with employees and managers to identify key inhibitors of employee–AI collaboration using thematic analysis conducted with MAXQDA. Study 2 examines how task complexity influences these inhibitors under different scenarios using MANCOVA. Study 3 tests the mechanisms linking inhibitors to collaboration willingness, with a focus on psychological empowerment, using mediation analysis based on Hayes’ PROCESS Model 4. Findings reveal that resistance to collaboration with AI is driven by four inhibitors: job insecurity concerns, AI complexity concerns, AI failure concerns, and privacy concerns. The results further show that task complexity strengthens these perceived inhibitors, especially when AI is completing complex tasks. Mediation analyses indicate that psychological empowerment plays a critical role in shaping collaboration willingness, with goal internalization and job autonomy emerging as the stronger mediators, whereas perceived capability has a weaker effect. Collectively, this dissertation contributes to theory and practice by offering a comprehensive framework of AI anthropomorphism and providing insights into how AI design and task characteristics influence employee well-being and collaboration in service environments.

UNDERSTANDING HUMAN-AI COLLABORATION – FROM ANTHROPOMORPHIC PERCEPTIONS TO WORKPLACE EXPERIENCES AND INTERVENTION STRATEGIES

Files and links (1)

Abstract

Metrics

Details