AI Augmentation and Employee Job Crafting: A Dual-Pathway Model of AI Self-Efficacy and Cognitive Fatigue
Description
As AI becomes increasingly integrated into employees’ work, AI augmentation is emerging as an important technological form that reshapes how employees perform their work. However, existing research has not yet fully explained the potential benefits and challenges that AI augmentation presents for employees. Drawing on job demands–resources theory, this study develops a dual-path model to examine how AI augmentation influences employees’ approach and avoidance job crafting through distinct psychological mechanisms and how AI literacy shapes the strength of these pathways. Data were collected from 337 employees in information technology and financial firms using a three-wave time-lagged survey. The results show that AI augmentation enhances employees’ AI self-efficacy, which in turn promotes approach job crafting. Conversely, AI augmentation may also increase employees’ cognitive fatigue, which in turn promotes avoidance job crafting. Furthermore, AI literacy strengthens the positive effect of AI augmentation on AI self-efficacy while weakening its positive effect on cognitive fatigue. The study further discusses the theoretical and practical implications of these findings and identifies directions for future research.
Files
Steps to reproduce
The sample consisted of employees from seven information technology and financial companies located in economically developed regions of China. To reduce the potential for common method bias (CMB), we followed the recommendation of Podsakoff et al. (2003) and adopted a three-wave, time-lagged design for data collection. Data were collected using online electronic questionnaires. Before the survey, participants were informed that their participation was voluntary and that the collected data would be used solely for academic research. Specifically, at Time 1 (T1), 573 questionnaires were distributed. Employees accessed the online questionnaire through WeChat and completed measures of AI augmentation, AI literacy, and demographic information. In addition, employees were asked to provide the last four digits of their mobile phone numbers solely to facilitate anonymous matching across the three waves. A total of 464 valid questionnaires were obtained at this stage (response rate = 80.98 %). One month later, at Time 2 (T2), the second wave of data collection was conducted among employees who had participated in T1, and they were asked to complete measures of AI self-efficacy and cognitive fatigue. A total of 389 valid questionnaires were obtained (response rate = 83.84 %). One month later, at Time 3 (T3), the third wave of data collection was conducted among employees who had participated in T2, and they were asked to complete measures of approach job crafting and avoidance job crafting. A total of 351 questionnaires were collected. After excluding invalid questionnaires and cases with missing data, 337 valid matched responses remained. Descriptive statistics and correlation analyses were conducted using SPSS 29.0. Confirmatory factor analysis (CFA) was conducted using Mplus 8.0 to assess the distinctiveness of the constructs. The measurement model was then assessed using SmartPLS 4.1, followed by hypothesis testing through partial least squares structural equation modeling (PLS-SEM).