Abstract: Given the widespread application of artificial intelligence (AI) technologies in workplaces, there has been a rapid increase in literature exploring AI-related themes. Scholars are increasingly focused on understanding how these applications influence employee behaviors and psychology. However, consensus on the direction, boundaries, and extent of these effects remains elusive. To address this issue, this paper conducts a meticulous review and selection of literature published from January 2017 to July 2023. A meta-analysis is performed on the 64 literatures (N = 150) to advance knowledge in three main areas: (1) Explore the strength and direction of the relationship between AI application and employees’ positive behaviors and psychological effects, as well as their negative behaviors and psychological effects. This aims to clarify the inconsistent conclusions and fill gaps in quantitative integration. (2) Based on the Job Demands-Resources model, this paper delineates the theoretical rationale underlying the impact of AI on employees’ behavior and psychology within an organizational context, upon its integration as a new technology, and elucidate specific pathways of its effects. (3) Investigate whether the effects of AI application on employee behavior and psychology are potentially influenced by the type of AI application, industry context, and measurement methods. Endeavor to furnish a clearer and more comprehensive overview of the correlation between AI and employee outcomes, thereby providing a theoretical foundation for tailored AI advantages in practical settings and methodological designs for subsequent empirical research in academia.
The result finds that: (1) The application of AI in the workplace exhibits a “double-edged sword” effect, which can enrich employees' psychological resources as technical support and stimulate positive behaviors, may also threaten employees to consume psychological resources and cause negative behaviors. (2) The relationships between AI application and employee behaviors/psychological effects vary under different AI types. Assisted and augmented AI enhance employee job satisfaction by reducing task costs, thereby increasing work engagement, creativity, and productivity. Such abundance in work resources contributes to an uplift in employees' job satisfaction and happiness. Consequently, when employees experience greater job involvement, there is a notable increase in creativity and productivity. However, managerial and autonomous AI types, despite improving efficiency and autonomy to some extent, introduce stress due to their supervisory and controlling attributes, suppressing positive work experiences and fostering negative psychological states. (3) Variations in AI application effects on employee behaviors and psychological effects across different industry types are evident. Employees in labor-intensive industries, with structured work environments and lower occupational skills, perceive more negative effects from AI. Conversely, employees in knowledge-intensive industries benefit from more flexible and autonomous work environments enhanced by AI, demonstrating stronger abilities in receiving, learning, and adapting to new information and technologies. (4) The relationship between AI application and employee behavior, as well as psychological impacts, varies depending on diverse measurement of AI application. Studies using subjective evaluations tend to reveal more negative impacts of AI on employee behaviors and psychological effects compared to those using objective measurement methods.
This study has made several theoretical contributions: (1) Systematically integrate and evaluate the fragmented research conclusions on the effects of AI application on employee behaviors and psychology, synthesizing empirical findings and responding to calls in the literature for understanding the personal impacts of automation technologies. (2) Within the framework of Job Demands-Resources model, this paper elucidates the diverse impacts of different types of AI application on employee behavior and psychology, expands the influencing factors that could augment the positive results of AI application, and further validates the concerns regarding potential adverse consequences. (3) Enrich the boundary conditions in the relationship between workplace AI application and employee behavior and psychology. This paper explores the moderating effects of the type of AI application, industry context, and measurement methods, responding to the scholarly calls for further examination of moderating variables of AI application affecting employee experience, thereby offering new insights for inconsistent research conclusions in the academic literature. Beyond theoretical advancements, the results of this study provide guidance for organizations to scientifically adjust the management strategies of AI, accurately direct employees perceptions, and effectively maximize its value.

Key words: workplace, AI application, positive behavior and psychology, negative behavior and psychology, meta-analysis