As GenAI-generated applications are increasingly deployed for employee counselling, organizations remain uncertain whether employees will engage with GenAI-generated counsellors and whether these investments will deliver organizational benefits. This research examines how employee evaluations of GenAI-generated counsellors shape perceived organizational attractiveness through perceived enjoyment, psychological safety and engagement willingness.
A three-stage multi-method design was used using a cross-sectional dataset of 311 employees. Stage 1 applied partial least squares structural equation modelling (PLS-SEM) to test the theory-driven structural model. Stage 2 employed elastic net and random forest machine-learning techniques to assess predictive robustness. Stage 3 applied Bayesian network analysis to examine whether the proposed directional sequence emerged from the data without being theoretically imposed.
Perceived enjoyment, psychological safety and engagement willingness serially mediated the effects of perceived anthropomorphism, adaptability, legitimacy, reliability and feeling heard on perceived organizational attractiveness. Perceived legitimacy emerged as the strongest predictor, while perceived enjoyment and engagement willingness showed the most stable effects across all analytical stages. Technostress weakened the relationship between perceived enjoyment and engagement willingness.
This research reframes GenAI counselling adoption as a question of employee evaluations under first-contact conditions. Drawing on dual-system theory, it shows that perceived enjoyment and psychological safety represent two mechanisms through which employee evaluations influence organizational attractiveness via engagement willingness. The study also demonstrates how a three-stage analytical design combining PLS-SEM, machine-learning and Bayesian network analysis can strengthen theoretical inference in organizational AI research.
