{"resourceId":"law-human-behavior-public-trust-human-hybrid-genai-judges-experiments-20261026","versions":[{"version":"external-081945427c915a7ea80f1aa87fba9287111de1281704a5bf707453d31af778b8","resource":{"id":"law-human-behavior-public-trust-human-hybrid-genai-judges-experiments-20261026","title":"Peer-reviewed experiments: the public trusts and accepts human judges and human-AI hybrids more than generative AI deciding alone, though results for low-involvement hybrids are mixed","organization":"Ludwig-Maximilians-Universität München (Department of Psychology); Law and Human Behavior (APA)","sector":"Courts / public acceptance of AI in judicial decisions","geography":"Germany (online experiments)","publishedAt":"Law and Human Behavior, 26 Oct. 2026 (PubMed record)","publicationDate":"2026-10-26","eventDate":null,"sourceName":"PubMed (Law and Human Behavior)","sourceLabel":"Judge v. generative artificial intelligence v. hybrid: The public trusts and accepts human and hybrid more than generative-artificial-intelligence-based judicial decision-making","sourceUrl":"https://pubmed.ncbi.nlm.nih.gov/42932807/","evidenceClass":"academic-research","outcomeClass":"mixed","topics":["knowledge-work","governance-procurement","operating-model"],"finding":"Three preregistered online experiments with German adults (N = 323, 674 and 772), drawn from convenience samples and panel providers, presented hypothetical crime cases and judgments attributed to a human judge, a hybrid with high or low human involvement, or generative AI. In Experiment 1, perceived ability, benevolence, integrity and acceptance fell as human involvement decreased. In Experiments 2 and 3, human and hybrid conditions were rated similarly, and both significantly higher than generative AI. Generative AI was seen as less able for emotional than technical crimes (d = 0.58). The authors conclude the evidence on hybrids is mixed but warns of losses in trust and acceptance if generative AI replaced rather than supported judges.","sledRelevance":"Gives court leaders evidence that how AI assistance is framed and how much human involvement is visible may affect public acceptance of judgments.","evidence":"Opened the PubMed record and read the full structured abstract; the publisher full text is closed access and was not read, so effect sizes beyond those in the abstract and exact manipulations are not verified.","architectureImplications":"Not an architecture study; implies workflow designs should keep the judge’s role visible and substantive.","governanceImplications":"Supports policies that keep AI in a supporting role and make human responsibility clear to parties.","securityPrivacyImplications":"Not addressed.","caveats":"Abstract only; hypothetical vignettes, not real litigants; German convenience and panel samples, mostly White, so transferability to U.S. court users is uncertain; public perception, not decision quality.","streamIds":["public-safety"],"roles":{"sales":"Sales framing: for court leaders the question is not only whether AI is accurate but whether litigants accept judgments it touches. This study suggests visible human involvement protects trust, while AI deciding alone loses it. A bounded engagement could test disclosure wording and human-involvement descriptions with a court’s own users. The evidence cannot support claims about decision accuracy, and its German online samples and hypothetical cases limit transfer to U.S. courts.","engineering":"Engineering framing: the design lesson is to keep human judgment visible in the workflow and the record. Pre-sales should check whether tools log who did what, so a court can truthfully describe human involvement in a disclosure. A proof of value would generate an audit trail for a sample of AI-assisted drafts showing the judge’s edits. Constraint: only the abstract was available, so experimental details beyond it are not verified.","delivery":"Delivery framing: the work is to pair any judicial AI rollout with tested public communication: disclosure language, clear statements of human responsibility, and user feedback. The owner is the court’s communications office with the court administrator. Acceptance criteria are proposals: disclosure text tested with court users before launch, and periodic surveys of litigant trust. Main risk: AI use that becomes known through controversy rather than disclosure, eroding acceptance."},"roleDetails":{"sales":{"customerProblem":"Courts adopting AI risk losing public acceptance if human responsibility is not visible.","stakeholders":["Chief judge and court administrator","Court communications office","Self-represented litigant services","Judicial ethics committee"],"discoveryQuestions":["How will the court describe human involvement when AI assists?","Has disclosure wording been tested with court users?","Which uses would look like AI deciding alone?"],"valueHypothesis":"Tested disclosure and visible human review could protect acceptance of AI-assisted judgments.","potentialEngagement":"User testing of judicial AI disclosure language and human-involvement descriptions.","unsupportedClaims":["Accuracy of AI-assisted judgments","Results for U.S. litigants","Effects in real cases"]},"engineering":{"fit":"Relevant to courts piloting drafting or summarisation tools.","architectureIntegration":"Audit trails in drafting tools that show judge edits and approvals.","prerequisites":["Edit and approval logging","Disclosure policy","User research capacity"],"constraints":["Vignette study only","German samples","Abstract-only review"],"security":"Protect logs that record judicial deliberation as confidential court records.","proofOfValue":["Produce an audit trail for sample AI-assisted drafts showing human edits."]},"delivery":{"work":"Draft and test disclosure language; survey court users periodically.","dependencies":["Judicial approval of disclosure","Communications staff","Survey tooling"],"ownership":"Court communications office with the court administrator","skillsAdoption":"Brief judges on how to describe AI assistance in orders.","governanceCheckpoints":["Ethics review of disclosure","User testing sign-off","Annual trust survey review"],"acceptanceCriteria":["Disclosure tested with at least 100 court users before launch","Annual survey of litigant trust","All AI-assisted orders carry approved disclosure where required"],"risks":["Undisclosed use revealed later","Over-claiming human involvement","Non-transferable findings"]}},"retrievedAt":"2026-11-01T02:53:00Z","enrichedAt":"2026-11-01T03:01:09Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Suggests communication with the public about judicial AI use matters for legitimacy.","procurementImplications":"Tools that automate decisions outright would face a legitimacy cost, per these hypothetical-case results.","operatingModelImplications":"Disclosure wording and human-involvement levels could be tested with court users before deployment.","peerReview":"peer-reviewed","sourceVerification":{"openedUrl":"https://pubmed.ncbi.nlm.nih.gov/42932807/","referenceExcerpt":"they warn of losses in trustworthiness and acceptance if GenAI replaced instead of supported judges but highlight the potential of hybrid judicial decision-making.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}