Purpose: The study explores whether the type of evaluator, human manager versus AI-based performance management system, affects employees’ intrinsic motivation, extrinsic motivation, perception of procedural justice and interpersonal justice in a non-Western entrepreneurial context. It addresses a gap in the AI-in-HRM literature, which has disproportionately studied large Western organizations and has left the motivational consequences of AI-based evaluation largely unexplored.
Method: A between-subject experiment was conducted with employees at a small business in Kathmandu Valley, Nepal (N = 31). Participants were assigned randomly to read a workplace scenario in which their performance evaluation was conducted by either a human manager or an AI-based performance system. Evaluation content and salary outcome were held constant across conditions. Dependent variables were measured using adapted items from the Motivation at Work Scale (Gagné et al., 2010) and Colquitt’s (2001) Organizational Justice Scale. Independent-samples Welch’s t-tests were used to test four directional hypotheses.
Findings: None of the four proposed hypotheses were supported. AI-condition participants reported higher intrinsic motivation than human-condition participants, with a medium effect size that approached conventional significance (d = 0.67, p = .068). Effects on extrinsic motivation, procedural justice, and interpersonal justice were statistically nonsignificant and negligible in magnitude (d ≤ 0.09). The intrinsic motivation reversal is interpreted through the lens of the algorithm appreciation framework and the autonomy-supportive properties of self-paced, non-interpersonal feedback delivery. Near-zero justice effects are attributed to outcome-dominance anchoring from the constant-outcome design.
Originality/Value: The study provides an experimental test of the effects of AI versus human evaluators on employee motivation and perceptions of justice in a non-Western startup context. The reversed intrinsic motivation finding challenges the foundational SDT-based assumption about the superiority of human-delivered feedback and contributes novel evidence to the algorithm appreciation literature. Findings are directly relevant to emerging regulatory debates about responsible AI deployment in the workplace across developing economies, where AI adoption is accelerating without commensurate oversight infrastructure
Keywords: performance appraisal, artificial intelligence, employee motivation, algorithmic evaluation, entrepreneurial ventures, Nepal