Abstract
In the context of performance appraisal, AI is increasingly being given the role of a decision aid – providing analytical inputs and recommendations, even though managers generally remain responsible for the final evaluation. By standardizing certain aspects of performance appraisals and providing data-driven insights, many argue that AI has the potential to reduce managerial bias and support decision-making. However, serious concerns remain about fairness in AI-driven appraisal systems. Importantly, we know little about how managers react when AI recommendations diverge from their own intuitions and initial judgments: how fair do they find AI suggestions, and do they tend to alter their own judgments in response?
Existing research typically adopts a universalistic stance, viewing fairness as a set of predetermined rules or principles that AI systems can either uphold or undermine. Drawing on Motivated Reasoning Theory, this study conceptualizes fairness not as an objective assessment, but as motivated judgments shaped by contextual factors and managers’ salient motives- recognizing that fairness is fundamentally “in the eye of the beholder.” Specifically, we propose that managers’ fairness perceptions of AI recommendations may be shaped by their empathy towards employees, the severity of the consequences of the appraisal decision, the directionality of the appraisal decision, and the salient judgment approach (evidence-based versus contextual understanding). These factors serve as informational cues that activate distinct motivational goals, thereby influencing perceptions of fairness. When the AI recommendation is more lenient (i.e., more positive) than managers’ initial performance recommendations, empathy may prompt managers to perceive AI recommendations as fair, making it more likely for them to adjust towards AI recommendations. In contrast, when AI suggests a more negative appraisal, and the consequences of performance decisions are meaningfully high, such as affecting promotion or career progression, empathy may drive managers to view AI as unfair. Similarly, the salient judgment approach to performance appraisal may affect managers’ fairness perceptions of AI and subsequent adjustments of their ratings, with more toward evidence-based practices supporting more positive perceptions of AI and stronger adjustments of one’s ratings