On The Administrative Regulation Of Algorithmic Automation Decision

After scientific and technological civilization entered the era of intelligence,artificial intelligent computers have similar functions as human brains,such as reasoning,decision making,learning,and so on,resulting in the concept of "automatic decision making algorithm".The automatic decision-making of algorithms goes beyond the role of human AIDS,and derives the ability of autonomous learning and deepening learning,so as to replace human beings to make decisions for specific problems or scenarios.The intelligence and concealment characteristics of algorithm automatic decision make the inner operation mechanism of algorithm decision difficult to be explained by human,which intensifies the opacity of algorithm automatic decision and causes the application risk of algorithm black box,algorithm discrimination and algorithm power.In this context,the Personal Information Protection Law and the Data Security Law are issued to ensure the security of personal information processing and data.The provisions on automatic decision-making in the law aim to respect human subjectivity and solve the functional problems of algorithmic decision-making,and initially reveal the meaning of algorithmic interpretation right and algorithmic rejection right.However,there are still three deficiencies in the automatic decision-making of administrative law regulation algorithm,including the absence of specialized review body,the influence of automatic decision-making on administrative due process,and the imperfect accountability mechanism of platform automatic decision-making.In order to realize the administrative law regulation of automatic algorithm decision-making,the following solutions are obtained by reading the relevant algorithm act of the European Union and the United States: First,administrative agencies should clarify the algorithm interpretation mechanism when making algorithmic decisions,so as to strengthen the transparency of algorithmic automatic decisions,and shape the ethical value of algorithms in the decision-making process,so as to reconcile the conflicts between algorithmic automatic decisions and administrative due process,and safeguard the digital rights and interests of citizens.Second,an algorithm safety committee should be set up to implement its institutional arrangement of "risk prevention and prudent supervision".At the same time,the power of the public and enterprises is introduced to strengthen the supervision of automatic decision-making.On the one hand,the digital rights of administrative counterparts should be established and the protection channels should be set up to improve the remedy of administrative counterparts’ rights.On the other hand,the special compliance management based on "algorithmic risk" promotes the autonomy of automatic decision-making of enterprise algorithms by combining algorithmic governance with enterprise compliance.Thirdly,the construction of institutional supervision should take the whole process of algorithmic decision-making as the basis,establish the pre-algorithm impact assessment system,the in-process algorithm audit system,and the post-algorithm responsibility division system,strengthen the whole-process management of technical supervision,and finally build a multi-level and refined supervision system.