Abstract: Algorithm interpretation is to make explanations to relevant subjects by means of visualization and natural language, while algorithm transparency is the state in which the subjects of algorithm obtain a certain degree of credibility and meet regulation through algorithm interpretation, and the relationship between the two is that of means and ends. The black box nature of the algorithm comes from its highly non-linear characteristics, but the law needs to condense the connection between the technical parameters of the algorithm and the “subjective malignancy” of the algorithm, and form the qualitative, quantitative and stereotypical standards for algorithm interpretation, so that the algorithm interpretation work can be standardized and understood by ordinary people. The “disclosure for monopoly” feature of the patent system makes it possible for the “disclosure” to meet the regulatory needs, while “monopoly” encourages enterprises to regulate themselves. Therefore, the theoretical evidence and institutional standards for judging the patent eligibility of artificial algorithms in China have gradually changed from “patentability” to “algorithm patent protection”. Under the background of regulation of algorithms covering different levels and categories, some tasks of algorithm transparency may be completed by algorithm patent disclosure documents, and at the same time, patent disclosure documents can become an indispensable basic reference in the subsequent algorithm transparency path.