Artificial Intelligence Evidence: Concepts, Nature and Applications

Abstract

Artificial intelligence (AI) has been widely applied in assisting judicial proceedings, but there remains some ambiguity and controversy regarding the use of machine opinions generated by AI as criminal evidence. AI evidence refers to materials formed through the use of AI to analyze and produce opinions such as voice recognition, image recognition, natural language processing, and expert systems, which are then used to prove case facts. It is characterized by generativeness, informativeness, and complexity. AI evidence is distinct from big data evidence and algorithmic evidence, as it manifests in the application of technologies like facial recognition, intelligent trajectory analysis, and generative AI in criminal cases. The admissibility of AI evidence can be judged based on its objectivity and legality, while its probative force can be analyzed in terms of relevance and admissibility. AI evidence should not be automatically categorized into existing statutory types of evidence, but rather, a miscellaneous provision could be added to the existing categories of statutory evidence. The review of AI evidence should involve improving the disclosure system and expert testimony rules, as well as drawing on the rules of corroboration, to ultimately promote a more scientific and widespread application of AI evidence in the judicial field.