Predictive Evidence in Criminal Proceedings: Attributes, Risks, and Applicable Rules

Predictive evidence refers to “algorithmic” similarity evidence that proves facts related to criminal procedure law in the form of algorithmic predictions based on data analysis, algorithmic models, and other technical means. In terms of its probative function and evidentiary characteristics, predictive evidence aligns with similar-fact evidence in traditional criminal proceedings. Although predictive evidence is not a statutorily recognized form of evidence, it can still serve as evidence to prove the facts of a case in certain circumstances. Meanwhile, the application of predictive evidence in criminal proceedings carries risks such as the transmission of cognitive biases, information obscuration, and interference with rights. These risks can easily lead to an imbalance between prosecution and defense in criminal litigation, as well as an imbalance in litigation values, ultimately resulting in the adjudicator making unfair judgments due to incorrect application of evidence. Therefore, it is essential to clarify the fundamental principles applicable to predictive evidence, and to construct operational rules for evidentiary capacity and probative force by combining the traditional rules of similar fact evidence with the characteristics of predictive evidence, in order to promote the standardization of the application of predictive evidence and achieve digital justice.