关键词： 人工智能, 算法规制, 算法制度主义, 算法权力

Abstract: In the digital era,artificial intelligence algorithms are being extensively applied across various social domains and have evolved into crucial auxiliary instruments for public administration.However,the distinctive technical properties of algorithms pose substantial legal challenges,rendering comprehensive legal regulation a scholarly consensus.Given that algorithm-induced legal challenges manifest across multiple dimensions,precisely defining the connotation of algorithmic regulation remains problematic.From the algorithmic institutionalism perspective,algorithmic regulation should elucidate the institutional dimension of algorithms while exploring their emergent institutional facets.Algorithmic regulation should be institution-oriented rather than goal-oriented.The merit of comprehending algorithmic regulation through algorithmic institutionalism lies in its capacity to achieve institutional integration across human-machine collaboration,value extension,and legal construction of algorithmic technology,thereby resolving the conceptual fluidity and structural complexity inherent in algorithmic regulation.Algorithmic regulation should encompass two primary components:first,restructuring algorithmic power by incorporating it into a reliable framework;second,establishing algorithmic norms by creating space for algorithmictechnology to develop unique social norms.The institutionalization of algorithmic power facilitates addressing the legal qualification challenges of algorithmic power while providing potential pathways to prevent its implicit expansion and alienation.Meanwhile,the exploration and construction of algorithmic norms can support addressing algorithmic interpretability challenges and optimizing the configuration of algorithm-related rights.Reconstructing algorithmic regulation through algorithmic institutionalism opens new potential directions for achieving algorithmic good governance.

Key words: artificial intelligence, algorithmic regulation, algorithmic institutionalism, algorithmic power