Stanford Study Finds AI Companies Increasing Recruitment Focused on Senior Positions
A study covering 41 countries found that companies adopting AI are overall increasing recruitment, but the new positions are mainly senior roles, while employment for junior positions is actually decreasing. Researchers concluded: "In occupations with high AI exposure, AI saves labor for junior employees but expands labor for senior employees."
The research was conducted by Bharat Chandar from Stanford University and Bouke Klein Teeselink from King's College London, with data spanning from January 2021 to March 2026. The sample includes 1.25 billion job postings and 154 million employment records provided by labor data company Revelio Labs, covering 41 countries including Brazil, Saudi Arabia, the UK, and the US. The study grouped companies based on AI adoption and compared personnel changes at different levels between the two groups.
The core result is that over five years, employment in senior positions at AI-adopting companies grew by 6.7%, while employment in junior positions decreased by about 2.5% to 3%. The proportion of junior employees in the total workforce declined by about 2 percentage points. Overall, the number of employees in companies grew by about 3.3%, but this was statistically only marginally significant. In terms of occupational structure, senior employees moved to AI-exposed positions by 0.9 percentage points, while junior employees moved out by 0.6 percentage points. The most impacted were computer and mathematics positions: these roles increased by 0.8 percentage points in AI companies, but the increase was almost entirely in senior levels, with the decline in junior proportions being the largest among all occupations.
Looking at individual countries, among the 31 countries with data, the proportion of junior employees declined in 23 countries, with significant declines in the US, Brazil, and Saudi Arabia. Data on young workers is even more telling: in AI-exposed occupations, employment levels for 22 to 25-year-olds are now 19% lower than expected, and this gap is widening.
Other data points in the same direction. Research from Hiring Lab, a subsidiary of job site Indeed, shows that in May 2026, job postings for senior positions increased by 14.7% year-on-year, while entry-level positions decreased by 7.5%. Venture capital firm SignalFire reported a 65% reduction in entry-level recruitment among 12 major tech companies since 2019. PwC's 2026 Global AI Employment Outlook found that among AI-exposed positions, 52% of newly emerging entry-level skill requirements involve "strategic judgment, leadership, and decision-making," abilities that were originally associated with senior employees are now being included in entry-level job requirements.
At the corporate level, KPMG UK reduced its graduate recruitment from 1,399 to 942, while Deloitte, EY, and PwC also reduced graduate hiring to varying degrees. There are also companies taking the opposite approach: IBM plans to triple its entry-level recruitment in the US by 2026, citing the need to maintain a long-term talent pipeline.
From a market mechanism perspective, this is a technology-driven adjustment of the labor structure. Companies, as buyers of labor, are shifting their labor budgets originally allocated for a large number of junior employees towards AI tool subscriptions, computing power, and a few high-paid senior employees. The flow of funds is: labor costs are converted into AI software and cloud service revenues, while simultaneously pushing up the salary premium for senior talent. Beneficiaries include AI model and enterprise software suppliers, experienced and judicious senior employees, and shareholders of companies that can expand output with fewer human resources. Those under pressure include fresh graduates, the outsourcing and professional services industries primarily focused on junior positions, and industries like consulting, accounting, and law that rely on a "pyramid structure" to cultivate talent for long-term supply.
Source: Public Information
ABAB AI Insight
Bharat Chandar previously collaborated with Erik Brynjolfsson, director of the Stanford Digital Economy Lab, to publish a paper titled "Canaries in the Coal Mine" in August 2025, which found that employment for 22 to 25-year-olds in AI-exposed occupations had relatively decreased by 13%, while employment for older workers in the same occupations remained stable or even grew. This new study expands the sample to 41 countries and 1.25 billion job postings, proving that this "young people are hit first" phenomenon is not unique to the US. Brynjolfsson proposed the concept of the "Turing Trap" as early as 2022, warning that if AI is primarily used to mimic and replace humans rather than enhance human capabilities, the benefits will concentrate in the hands of capital owners.
From a capital pathway perspective, large enterprises are directly converting labor budgets into AI budgets. In June 2025, Amazon CEO Andy Jassy stated in a letter to employees that AI would reduce the company's white-collar workforce in the coming years, and in October of the same year, Amazon cut about 14,000 corporate positions. In September 2025, Salesforce CEO Marc Benioff mentioned that the company had reduced about 4,000 customer service positions due to AI customer service agents. Microsoft laid off about 15,000 employees in 2025 while increasing its annual capital expenditure to over $80 billion, mainly for AI data centers. The flow of funds is shifting from payroll to computing power, with the first jobs to be compressed being standardized and replicable junior roles.
In historical analogy, economists refer to similar phenomena as "skill-biased technological change." During the 1980s and 1990s, when computers became widespread, the wage premium for high-skilled employees rose significantly. MIT economist David Autor found that routine jobs requiring medium skills were largely replaced, leading to a "polarization" in the labor market. Another analogy is ATMs: when ATMs became popular in the 1970s, economist James Bessen's research found that while the number of tellers at individual branches decreased, banks continued to grow the total number of tellers through branch expansion, shifting job content towards sales and customer relations. The difference with this round of AI is that it primarily replaces cognitive white-collar jobs, impacting the very bottom rung of the career ladder.
Essentially, this is a technological substitution, but the objects of substitution are "codifiable knowledge." The mechanism is that the primary value of junior employees comes from explicit knowledge learned from books that can be codified into rules, such as writing basic code, organizing data, and drafting standard documents, which is exactly what large models excel at; the value of senior employees comes from years of accumulated tacit knowledge, such as prioritizing tasks, handling exceptions, and managing client relationships, where AI acts as an amplifier rather than a substitute. Thus, AI simultaneously reduces the demand for junior labor and increases the productivity and bargaining power of senior labor. The long-term risk lies in the talent supply chain: today's senior employees have all grown from junior positions, and if companies generally stop hiring newcomers, there will be a gap in senior talent capable of managing AI in ten years, which is precisely the logic behind IBM's counter-cyclical expansion of entry-level recruitment.
ABAB News · Cognitive Law
AI first replaces knowledge written in books, then elevates experience not documented in books.
Companies cutting apprentice positions are over-drafting future experts.
The efficiency dividend first goes to capital, then to seniors, and only then to newcomers.