Abstract: Algorithmic decisions driven by artificial intelligence are subject to the combined influence of multiple factors, including the inherent logic of the algorithms, human bounded rationality, and user feedback loops. This has led to issues such as insufficient representation of specific social groups, high error rates in identification, and biases in policy applicability. In response, the government, market, and social actors have all made certain efforts to address these issues, exploring into technical logic, ethical principles, and application limitations. However, the breadth, depth, and sustainability of these efforts are relatively limited. To build a public governance system that addresses algorithmic bias, it is necessary to promote collaborative interactions among the government, market, and various social stakeholders, facilitating the complementary embeddedness of bureaucratic, market and community mechanisms. The government must shift from merely relying on bureaucratic mechanisms to performing meta-governance functions. By establishing a public governance system centered on collaborative governance and shared responsibility among multiple stakeholders, and by institutionalizing the public governance of algorithmic bias, we can create new productive relationships and achieve the symbiotic development of technological transformation and social order.