Michelle Caswell and others have applied critical theory to the field of archival studies, forming the framework of "critical archival studies", which emphasizes a broad focus on the interests and cultural contexts of power embedded in archives. This framework offers new perspectives and methodologies for the development of archival studies in the AI era, becoming an academic starting point for critical and reflective perspectives on archives. The application of digital technologies such as machine learning and large language models has profoundly influenced social development. Research confined to archives themselves is no longer sufficient to address the challenges posed by the era of artificial intelligence, necessitating reflection and transformation. Currently, the authenticity of archives faces dual challenges from human society and algorithmic power. The memorial function of archives is further weakened by the technical limitations of archival practitioners and the intervention of algorithmic entities. Additionally, the increasing diversity of archival sources in the context of human-AI collaboration has introduced broader forms of oppression and bias. In response, archival practitioners must rethink and redefine the concept of "archives" in the AI era, develop a deeper understanding of and exercise effective control over the influence of algorithms on archives, uphold the central role of humans in the construction of digital memory, and actively contribute to the efforts of philosophy and social sciences to uphold digital fairness and justice.