THE ROLE OF ARTIFICIAL INTELLIGENCE IN CRIMINAL PROCEDURE: EVIDENCE EVALUATION AND LEGAL RESPONSIBILITY

Abstract

Abstract: The integration of artificial intelligence (AI) into the criminal justice system has changed traditional methods of investigation, evidence evaluation, and judicial decision-making. Artificial intelligence-driven tools are increasingly used in the fields of predictive police, facial recognition, digital forensics, and risk assessment. Although these technologies provide efficiency and analytical advantages, they have also aroused deep legal concerns about evidence admissibility, procedural fairness, and legal liability. This paper examines the evidentiary value of the results produced by artificial intelligence and analyzes the distribution of legal responsibilities arising from artificial intelligence decision-making in the criminal procedure Law.In terms of risk control, in order to prevent the “illusion” of large legal models, the technical design has set strict “generation boundaries”, such as dismantling tasks into three types of retrieval, reasoning, and generation capabilities, eliminating all aspects of a single model processing, and prohibiting fabricated content in high-risk scenarios such as case citations. It must be retrieved from authoritative databases, while ensuring that the reasoning process can be traced, positioning AI as a “capability amplifier”, rather than replacing the core discretion of judicial personnel. It believes that if there are no clear legal standards on transparency, reliability, and accountability, the use of artificial intelligence may undermine the basic principles of criminal justice. This article proposes regulatory and theoretical methods to ensure that AI becomes a supportive rather than decisive tool in criminal proceedings.