Figure 3
The diagram presents the alternative collective FAT model. Interaction quality, human agency, perceived humanness, task complexity, and perceived uncanniness are modelled as antecedents of overall FAT judgement. Interaction quality is positively related to overall FAT judgement (β = 0.189), as are human agency (β = 0.348), perceived humanness (β = 0.361), and task complexity (β = 0.158). Perceived uncanniness has a small negative, non-significant relationship with overall FAT judgement (β = −0.093). Overall FAT judgement represents the three dimensions of fairness, accountability, and transparency, with respective weights of 0.362, 0.329, and 0.445. Overall FAT judgement is strongly and positively related to trust in AI-supported performance evaluation (β = 0.781). The model explains 71.9% of the variance in overall FAT judgement and 61.0% of the variance in trust. Solid paths indicate statistically significant relationships, while the dashed path indicates a non-significant relationship.Alternative collective FAT model. Source: Authors' own work