Judge v. generative artificial intelligence v. hybrid: The public trusts and accepts human and hybrid more than generative-artificial-intelligence-based judicial decision-making.

OBJECTIVE: Generative artificial intelligence (GenAI) could transform judicial decision-making by assisting or replacing judges. Prior research indicates that laypeople trust and accept human judges more than algorithmic judges. However, little is known about public perceptions of hybrid decision-making-human judges with GenAI support-which will represent the most realistic scenario in future judiciaries worldwide. HYPOTHESES: Building on the organizational trust model, machine heuristic, and Capability-Personalization Framework, we investigated the effect of the type of decision-maker on perceived trustworthiness (ability, benevolence, and integrity) of the decision-maker and acceptance of judgments. We hypothesized higher ability, benevolence, and acceptance and lower integrity for human and hybrid judicial decision-making than for GenAI. Additionally, we postulated lower ability and acceptance for GenAI for emotional than for technical crimes. METHOD: Across three preregistered online experiments with German adults (most identified as White) from convenience sampling and panel providers (PsyWeb, Prolific), we presented hypothetical crime cases, court proceedings, and judgments. We manipulated the type of decision-maker as between factor with four levels: (a) human judge, (b) hybrid with high human involvement, (c) hybrid with low human involvement, and (d) GenAI. Participants answered questionnaires assessing the decision-maker's trustworthiness and the acceptance of judgment. CONCLUSIONS: Findings provide mixed evidence on the perception of hybrid judicial decision-making but support preferences for human judges. Overall, they warn of losses in trustworthiness and acceptance if GenAI replaced instead of supported judges but highlight the potential of hybrid judicial decision-making. (PsycInfo Database Record (c) 2026 APA, all rights reserved).