The Impact of Generative AI on the Job Market

Recently, two Harvard University researchers, Seyed M. Hosseini and Guy Lichtinger, published a paper titled "Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data." By analyzing data from 2015-2025 covering nearly 285,000 U.S. companies, 62 million employees, and 245 million job postings, they concluded that since early 2023, generative AI has been reshaping the labor market in a "seniority-biased" manner, with far greater impact on junior employees than senior ones.

The Harvard study reveals that companies adopting AI saw a relative 7.7% decline in junior employee numbers over six quarters; senior positions were largely unaffected and even showed slight growth; this phenomenon was primarily driven by reduced hiring rather than mass layoffs; the wholesale and retail sectors were most affected, with junior hiring dropping by 40%.

This divergence stems from AI's efficient replacement of repetitive tasks. Basic code debugging, preliminary legal document reviews, and business document drafting—traditional entry-level work—can now be completed more efficiently by AI, significantly reducing corporate demand for entry-level newcomers.

The study categorized employee university backgrounds into five tiers (Tier 1-5), from top elite institutions to regional universities. The research showed AI's impact on graduates follows a U-shaped pattern, with Tier 2-3 (mid-tier institutions) graduates experiencing the greatest impact, while Tier 1 (top-tier) and Tier 4-5 (lower-tier) graduates faced relatively less impact.

The study also found that within AI-adopting companies, promotion rates for existing junior employees increased, indicating companies prefer internal advancement over external hiring.

AI Forces Young People to Extend Career Preparation, Delaying Marriage and Childbearing

AI development objectively widens the income gap between high-skilled and low-skilled workers. High-skilled AI senior employees are even reaching record-high incomes, but reduced junior positions make employment more difficult for college graduates. Acquiring high skills requires higher education and more work experience. The traditional path of "employment immediately after bachelor's degree" is gradually becoming obsolete. To break through competitive bottlenecks, more bachelor's graduates choose to pursue master's and doctoral degrees or participate in more internships. These additional qualifications and internship experiences typically offer no income or very low income, requiring years of accumulation before finally entering high-income brackets.

In recent years, domestic graduate enrollment has increased dramatically, with some universities enrolling more graduate students than undergraduates. For example, Nanjing University's 2025 incoming class includes 4,113 undergraduates and 9,222 graduate students—more than double the undergraduate number. Fudan University enrolled 4,000 undergraduates and 12,000 graduate students in 2025, with graduate students being three times the number of undergraduates.

This phenomenon has some rationality, given that average lifespan and retirement age are both extending. However, under this influence, young people experience an extended period of meager income (possibly extending into their twenties or even thirties), leaving them without sufficient economic capacity to start families. Yet optimal childbearing age (especially for women) doesn't extend correspondingly with lifespan. This delayed employment trend means that if young people want to marry and have children before age 30, it conflicts with their educational and career plans. Without adequate income for marriage and childbearing, many young people must postpone these life events, with some choosing to remain unmarried and childless. Longer educational journeys also mean education expenses comprise a higher proportion of family spending, further raising modern society's educational costs—a concern that significantly contributes to declining fertility rates.

China's fertility rate in recent years has been only about half the replacement level, with 2023's total fertility rate at just 1.01—less than half the replacement level. At current fertility rates, the birth population will shrink by half every generation (30 years). This trend will sequentially cause severe negative impacts across all industries: first infant formula, children's products, and childcare services; then education, food, and clothing; followed by housing, furniture, home appliances, consumer electronics, automobiles, tourism, and entertainment; finally healthcare, elderly care, and funeral services. The impact on consumer-facing industries will gradually transmit to business-facing sectors. Moreover, these effects won't just cause actual demand contraction but will continuously depress expected demand, leading to sluggish domestic investment sentiment.

Population Size Becomes More Important in the AI Era

Some believe that as AI replaces more jobs in the AI era, China won't need such a large population, making it unnecessary to increase fertility rates.

I believe population size becomes even more important in the AI era.

First, innovation becomes more crucial in the AI era, as non-innovative work is primarily handled by AI, while AI systems become increasingly complex. AI system iteration and upgrades require continuous streams of innovative talent: from breakthrough algorithms and data model optimization to deep AI integration with various industries (industrial internet, intelligent healthcare, autonomous driving, etc.), each segment requires top research talent, versatile application specialists, and numerous professionals with relevant skills. Cultivating and maintaining this talent pool requires a sufficiently large population base as a "talent reservoir."

Larger population bases increase the probability of producing people with various professional skills and innovative capabilities. Different individuals possess different talents, interests, and educational backgrounds; within larger population scales, all industries can produce abundant excellent talent. For instance, in information technology, populous countries like India have cultivated large numbers of outstanding software engineers.

In internet and artificial intelligence fields, users are not only product consumers but innovation participants. With search engines, for example, user search operations generate data on search keywords and click behaviors. This data optimizes algorithms, enabling search engines to better understand user intent and provide more relevant results. Similarly, on social platforms, user-uploaded content and interactions enrich platform content ecosystems and prompt platforms to improve functionality based on user feedback. Therefore, increasing user numbers provides more data support for product algorithms, making them more user-friendly; from a content perspective, more users generate more content, creating a snowball effect.

AI industry competition is essentially "scale competition"—only countries simultaneously possessing massive population scales, abundant innovative talent, and comprehensive industrial categories can dominate this competition. Currently, very few countries globally possess independent AI research and industrialization capabilities. China's position as a major player benefits from unique scale advantages: a 1.4 billion population forming an ultra-large-scale market provides rich application scenarios and massive data resources for AI technology; a huge young population supports talent needs across the entire industrial chain from chip manufacturing and software development to industry applications; comprehensive industrial categories enable AI technology to rapidly penetrate various fields, forming a virtuous cycle of "technology R&D—industry application—data feedback—technology upgrade."

However, China's fertility rates and birth population continue declining in recent years, population aging accelerates, and gradual shrinkage of young population scales may fundamentally weaken the talent foundation and market vitality for AI industry development. Without timely fertility rate improvements, China's AI competitiveness may gradually decline with young population depletion—when R&D talent gaps appear and market scale growth slows, AI technology iteration momentum and industrial competitiveness become difficult to maintain.

Solutions: Provide Birth and Education Subsidies to Young People

Young people face personal financial difficulties during three stages: education, marriage/childbearing, and early career phases. Current society, based on technological and AI development, fully possesses the capability to allocate resources for subsidizing education, fertility, and internship training.

Recently, the State Council General Office issued "Opinions on Gradually Implementing Free Preschool Education," clearly stating that starting from the autumn 2025 semester, childcare and education fees will be waived for children in their final preschool year at public kindergartens. This represents welcome progress but remains insufficient. We recommend rapidly implementing complete free preschool education to genuinely reduce educational burdens on child-rearing families. Three years of free preschool education combined with existing nine years of free compulsory primary and secondary education totals twelve years of free education. Future expansion could include free high school education.

Cash subsidies for childbearing have proven effective through international and domestic experience, provided the magnitude is sufficiently large. If birth subsidies lack effectiveness, it's because subsidy levels are too small. According to calculations, significantly boosting fertility rates requires birth subsidy fiscal expenditure of at least 2%-5% of GDP. Specific birth subsidy recommendations are detailed in the appendix.

Some worry whether such large-scale cash distribution to young people might trigger inflation. Reaching conclusions on this requires examining whether production capacity has surplus capacity and whether employment is sufficient. China currently faces overcapacity and insufficient employment, so expanding consumption can utilize idle capacity and labor. China currently faces not inflation but negative price growth, precisely needing deficit fiscal stimulation. Naturally, deficit spending cannot become a long-term solution and can be appropriately adjusted during inflationary cycles. According to basic economic principles, fiscal deficit spending has obvious multiplier effects under insufficient demand conditions without triggering inflation. Therefore, using fiscal deficits for childcare subsidies will have significant economic expansion effects, proving faster than interest rate cuts and other stimulus policies while maintaining long-term innovation vitality.

China currently has ample production capacity in all sectors; the only shortage is children. This situation results from investment-return mismatches. Child-rearing requires family investment, but returns from children's growth are enjoyed by entire society. This problem must now be addressed at the national level through central fiscal policies providing more benefits to child-rearing families. Providing money to child-rearing families directly reaches the real economy through household consumption while promoting employment and forming economic growth expectations.

Conclusion:

The greatest imbalance in the AI era is that while AI drives total social wealth growth through improved production efficiency, new industry creation, and optimized resource allocation, AI simultaneously replaces junior intellectual labor, exacerbating difficulties young people face during education, marriage/childbearing, and early career stages. Government subsidies of significant magnitude for young people are necessary to resolve economic imbalances in the AI era.

Appendix: Specific Birth Subsidy Measures

Specific policy recommendations for birth subsidies include:

First, cash subsidies: Provide monthly subsidies of 1,000 yuan for each first child, 2,000 yuan for each second child, and 3,000 yuan for each third child and above, until children reach age 16 or 18.

Second, personal income tax and social insurance reductions: For two-child families, implement 50% personal income tax and social insurance reductions; for three-child and above families, completely waive personal income tax and social insurance (with subsidy caps for particularly wealthy families).

Third, housing purchase subsidies: Implement subsidies through mortgage interest returns. For example, subsidize 50% of mortgage interest for two-child families and fully subsidize mortgage interest for three-child and above families (not exceeding subsidy caps).

Related Article Link: Seyed M. Hosseini & Guy Lichtinger: Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5425555