Against Zeleznikow's warning that prediction is not judgment, this article proposes a duty to explain, a pre-deployment legal impact assessment, and an audit standard for fair process to keep judgment human.
Abstract
Algorithms have moved from the filing room to the bench. Machine learning, large language models, and online dispute resolution now shape bail, sentencing, and pleadings faster than the law can settle what this means for a fair hearing. Examining five jurisdictions across four continents, 2015–2025—47 statutes and judgments, 142 studies, seven datasets—this article finds three faults. Proprietary models and hidden training data leave the party bound by a decision unable to see its reasoning. Prediction displaces explanation: what usually happens stands in for why this case is decided. Risk falls on the litigants least able to absorb error, not on the institutions deploying the systems. Against Zeleznikow’s warning that prediction is not judgment, it proposes a duty to explain, a pre-deployment legal impact assessment, and an audit standard for fair process. The aim is not to keep artificial intelligence out of the courtroom, but to keep judgment human.
This work identifies the 'inertia of confidence'--an overconfidence phenomenon analogous to the Dunning-Kruger effect where LLMs provide incorrect legal verdicts with near-maximum confidence, driven by a hypothesized'precedent overfitting'bias and proposes shifting toward adversarial legal research pedagogy and impleme...
Angel Mary John, Vipin Kumar Singh, J. T. Panachakel· 0 citations
It is found that at least for the time being, explicit normative instructions are not fully able to realign AI advice with the normative convictions of the population, or the legislator deciding on its behalf.
This article reports on a follow-up investigation to a 2023 study examining the academic performance of Generative AI (GenAI) in university-level Criminal Law assessment. We evaluated GenAI models from five artificial intelligence (AI) providers across two compulsory subjects in the Bachelor of Laws (LLB) degree at the...
Armin Alimardani, Melissa D. Porter, W. Gullett· Law, Technology and Humans· 0 citations
The unlicensed use of copyrighted works to train generative artificial intelligence (AI) systems is among the most contested questions in contemporary copyright law. Courts and scholars remain divided on how the fourth fair use factor should apply to a process spanning data ingestion, model training, and commercial dep...
This study studies how large language model (LLM)-powered chatbots respond to series of questions about legal reasonableness, and suggests that chatbot responses generally track those of human participants.
Nirav Patel, Emily Wenger, C. Buccafusco· Proceedings of the AAAI/ACM...· 0 citations
OBJECTION, an inference-time pipeline that integrates an Adversarial Lawyer Agent into each 3-step reasoning of offense, unlawfulness, and culpability, denotes a key progress toward aligning Legal AI with the presumption of innocence.
Jaehoon Jeong, Jay-Yoon Lee· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
