Liu Bingni (Southeast University) has posted Pathways to Judicial Justice in Legal Large Language Model-assisted Sentencing: A Legal Realist Perspective on SSRN. Here is the abstract:
The legitimate boundaries of legal large language model (LLM)-assisted sentencing essentially delineate the equilibrium between technical rationality and judicial rationality, as well as between legal realist practice and normative ideals. Although LLM-assisted sentencing ostensibly aligns with the legal formalist pursuit of rule deduction and legal certainty, its data-driven nature reveals a distinctly legal realist orientation. This orientation, in turn, engenders a dual crisis of judicial justice: a legitimacy crisis concerning fact-finding, and, at a deeper level, a normative crisis regarding rule application. Addressing these intertwined crises necessitates the establishment of virtue-centred human-machine collaborative accountability rules, an algorithmic governance framework that incorporates broad participation from the legal professional community, and a paradigm shift in judicial decision-making from an efficiency-first approach to one guided by virtue. This article seeks to provide jurisprudential support and institutional principles for the reform of sentencing standardisation in the digital age.
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum