Legal Regulation of Algorithms From the Perspective of Interpretability

Abstract

The human life in the age of artificial intelligence has undergone tremendous changes. Algorithm technology is widely developed and applied as one of the core technologies of artificial intelligence. However, a series of problems such as algorithm discrimination, algorithm killing, and “information cocoon” caused by unexplainable algorithms represented by artificial neural networks needed to be solved urgently,which forms a risk society. The algorithmic order is gradually “offside” into a new social order, which challenges the existing legal order. Because the existing legal order upholds the neutral value of technology tools and does not pay attention to the legal regulations of technology itself, it cannot make ethical prejudgment of unexplainable algorithms to prevent and control social risks. With the deepening of the “intelligence” of algorithm technology, social risk is expanding. The field of algorithm technology creates interpretable algorithms to respond to social risks. However, due to the lack of legal value and institutional design support, the technological advantages of interpretable algorithms to prevent and control algorithm black boxes, “Offside order” and coping with a risky society cannot be confirmed and guided by law. Therefore, it is an effective way to solve the risks in the age of artificial intelligence by taking the technical critical theory of risk society as the value basis and taking the interpretability of algorithms as a necessary condition for algorithm regulation.