关键词: 人机协作, 人工智能, 工作绩效, 工作幸福感

Abstract: As artificial intelligence (AI) becomes deeply embedded in organizational contexts, employee-AI collaboration has emerged as a prevalent work pattern. Despite its growing importance, existing research reveals three critical gaps. First, the conceptualization of employee-AI collaboration remains fragmented, lacking an integrated typology and validated measurement instruments. Second, prior studies predominantly examine the net effects of single factor on employee-AI collaboration, neglecting the configurational nature of how multiple factors jointly shape employee-AI collaboration patterns. Third, the consequences of employee-AI collaboration yield inconsistent findings, with limited attention to differentiating collaboration types and the dual cognitive-affective mechanisms linking them to employee outcomes. To address these gaps, this research develops a comprehensive framework, and conducts three studies encompassing the typologies, antecedents, and consequences of employee-AI collaboration.
Study 1: Employee-AI Typologies and Scale Development.
Study 1 proposes a 2 × 2 typology of employee-AI collaboration based on two fundamental dimensions: agency (employee-centric vs. AI-centric) and interaction intensity (low vs. high). This dual-dimensional framework moves beyond prior unidimensional classifications and captures the complexity of human-AI partnering into four typologies: Augmentation, Symbiosis, Assistance and Substitution. Below are the detailed explanations of four typologies.
Augmentation employee-AI collaboration (employee-centric agency, high interaction): Employees remain core actors while AI serves as an intelligent partner, providing knowledge transfer and decision support through frequent bidirectional interactions to enhance employees’ professional capabilities.
Symbiosis employee-AI collaboration (AI-centric agency, high interaction): Employee and AI form a tightly coupled joint cognitive system, mutually dependent and complementary, co-optimizing tasks and decisions through real-time adaptation.
Assistance employee-AI collaboration (employee-centric agency, low interaction): AI acts as a passive tool responding to specific commands without engaging in employees’ cognitive processes or capability development.
Substitution employee-AI collaboration (AI-centric agency, low interaction): AI autonomously performs tasks originally done by employees, who retain minimal supervisory control.
Based on the above four typologies, Study1 develops and validates four separate scales corresponding to each collaboration type.
Study 2: Configurational Antecedents of Employee-AI Collaboration.
Grounded in sociotechnical systems theory, Study2 identifies eight contextual factors spanning four subsystems: employee (employee’s AI literacy, employee’s AI awareness), AI (AI capability, AI reliability), task (task complexity, task type), and organization (organizational resources, organizational culture). Using fuzzy-set qualitative comparative analysis (fsQCA), Study2 investigates their synergistic effects through a configurational approach, and uncover how combinations of these factors jointly lead to different collaboration patterns.
Study 3: Consequence Mechanisms and Interventions of Employee-AI Collaboration.
Based on the Cognitive-Affective Processing System (CAPS) theory, study 3 proposes a dual-pathway model linking each collaboration type to employee job performance and well-being through cognitive (cognitive expansion, cognitive slackness) and affective (positive affect, negative affect) mechanisms. This study also theorizes four matched managerial interventions: decision sovereignty cultivation for Augmentation (reinforcing employee agency), dynamic contribution evaluation for Symbiosis (ensuring perceived fairness), function expansion programs for Assistance (preventing instrumental thinking inertia), and transition support for Substitution (mitigating replacement anxiety). Hypotheses include linear, curvilinear, and comparative effects across types.
Consequently, this research makes several important theoretical contributions in employee-AI collaboration field. First, this research develops an integrative typology of employee-AI collaboration. By integrating agency and interaction intensity, this research constructs a 2×2 framework categorizing collaboration into four distinct typologies: Augmentation, Symbiosis, Assistance, and Substitution. This dual-dimensional approach overcomes the limitations of prior unidimensional classifications and provides conceptual clarity. It further develops and validates corresponding scales, addressing the lack of measurement instruments and enabling future empirical research.
Second, adopting a configurational approach, this research reveals how employee, AI, task, and organizational factors interdependently shape employee-AI collaboration patterns. It advances socio-technical systems theory by specifying how social and technical subsystem elements combine to produce distinct work system configurations in the AI era.
Third, this research unpacks the cognitive-affective mechanisms through which different collaboration types affect employee outcomes, addressing calls for more comprehensive process models in human-AI research. It explains why collaboration can be both beneficial and detrimental—depending on collaboration type and mediating pathways—and responds to the neglect of affective processes in prior work. Finally, introducing corresponding managerial interventions offers actionable insights for organizations seeking to optimize employee-AI partnerships. By specifying which interventions fit which collaboration patterns, the research provides a contingency framework for human resource management in increasingly hybrid work environments.
The above findings will help employees recognize their collaboration patterns and develop complementary skills; guide managers in tailoring AI implementation and support strategies; and enable organizations to allocate resources effectively toward interventions that enhance both performance and employee well-being. The ultimate goal of this research is to foster human-AI synergy where technology augments rather than diminishes human potential at work.

Key words: Human-AI collaboration, AI, job performance, job well-being