Abstract: With the rapid development of artificial intelligence, the substitution risk of cognitive occupations such as copywriting, programming, and accounting has become increasingly prominent, and the global labor market is undergoing structural adjustment. This paper analyzes the different channels through which general-purpose and specialized large models affect jobs and income; compares the advantages, disadvantages, and applicable boundaries of four measurement methods, namely theoretical task exposure, observed indicators, firm-level micro evidence, and macroeconomic regression; reviews the negative psychological transmission loop formed by long-term reliance on AI, and analyzes the differentiated attitudes of young people, women, and workers in different occupations; summarizes four areas of consensus and four categories of core controversies in the literature, and proposes future research directions such as multidisciplinary integration and localized measurement; and provides targeted policy suggestions based on China's industrial characteristics.