The algorithms of social media platforms may exacerbate social divisions by filtering bubbles of information, strengthening extreme views, creating group antagonisms, and limiting information diversity. The details are as follows： -**Formation of information filtering bubbles**: The algorithm will accurately push content that meets user preferences based on user interests, browsing history and interactive behavior. This makes the information that users come into contact with more and more limited to their existing views and interests, as if trapped in an “information filtering bubble”. -**Strengthen extreme views**: Algorithms tend to push content that can arouse strong emotions and high attention to attract users' attention and increase user stay time. Extreme, one-sided, or emotional views are often easier to get a lot of attention, likes, comments, and sharing, resulting in more recommendations from algorithms, forming a vicious circle that further exacerbates the polarization of views. -**Create group opposition**: Through precise push, the algorithm makes it more difficult to communicate and understand between groups of different viewpoints. When different groups are only exposed to information that is consistent with their own views, they will gradually strengthen their sense of identity with their own groups, while deepening their misunderstandings and prejudices against other groups, leading to the escalation of antagonistic emotions between groups, and ultimately exacerbating social division. -**Limit the diversity of information**: The algorithm recommendation mechanism makes it difficult for those niche but valuable opinions and information to obtain sufficient exposure, limits the possibility of public access to diverse opinions and information, hinders mutual understanding and tolerance between different groups, and is not conducive to social harmony and unity.