This study uses a corpus generated by ChatGPT to apply cognitive linguistics, philosophy of technology, and sociolinguistics in analyzing how generative AI impacts the Chinese lexical system. Grounded in the “replication-variation-selection” model of language evolution, the research finds that abstract concepts created by AI mimic human conceptual systems via “algorithmic metaphors,” yet lack experiential grounding. Platform algorithms restructure language dissemination, establishing a “digital register” communication hierarchy, while the technological framework leads to a vocabulary hollowing effect, necessitating a new “algorithm governance” dimension in language planning. The study offers a theoretical framework for human-machine language interaction, suggests AI language ethics policies, and redefines traditional language evolution models by proposing a three-stage process: “variation-algorithm screening-platform diffusion.” It extends meme theory, identifying “technological memes” as engineering-driven replications. Practically, it provides strategies for language planning by advocating transparency, diversity, and morality as core algorithm planning principles, helping to address language alienation, maintain a dynamic balance between innovation and standardization, and support effective communication and the healthy evolution of language in the age of generative AI.