Mitigating algorithm aversion in the judiciary: How systemic dissatisfaction drives AI acceptance

Artificial intelligence (AI) algorithms are increasingly proposed to enhance efficiency and consistency in judicial decision-making. However, algorithm aversion—the general resistance to automated decisions—presents a significant challenge to their acceptance. Here we investigate whether dissatisfaction with the human legal system mitigates this aversion. Using data from the nationally representative 2023 Taiwan Law and Social Change Survey (N=1,660–1,883), we examine how perceptions of substantive fairness, procedural satisfaction, and actual courtroom experience influence the acceptance of AI judges. We find a clear hierarchy of acceptance: the public strongly prefers AI in consumer contexts over judicial ones, and favors civil applications over criminal ones. Crucially, we identify specific predictors of AI acceptance: individuals who perceive human judges as having poor demeanor are significantly more willing to accept AI decision-making. Nuancing the substitution hypothesis, however, we observe an asymmetry in direct experience: negative courtroom experience did not drive support for AI; instead, positive experience actively reinforced algorithmic aversion in criminal context. These findings suggest that while specific human shortcomings can facilitate preference for algorithms, the perceived value of human agency creates a resilient "human–AI gap" that technology alone cannot bridge. • Support for judicial AI declines as the task's moral gravity increases. • Perceived legal unfairness drives the preference for algorithmic substitution. • Dissatisfaction with judicial demeanor significantly predicts AI acceptance. • Positive courtroom experience hardens resistance to AI, particularly in criminal contexts. • The substitution effect is driven by interpersonal, not procedural, dissatisfaction.