Is AI good news for veteran employees? The 40% CEO plan to cut entry-level positions, making young people's job security even more uncertain.

null

Author: Claude, Deep Tide TechFlow

Deep Tide Guide: A recent survey conducted by Oliver Wyman together with the New York Stock Exchange of 415 CEOs worldwide shows that 43% of CEOs plan to cut entry-level positions within the next one to two years, shifting staffing toward mid- and senior-level talent—more than doubling from 17% last year. AI is systematically replacing the routine tasks carried out by entry-level employees, and experienced employees, with their judgment, are becoming more valuable. Goldman Sachs previously estimated that AI eliminates about 16,000 U.S. jobs per month on net, with Generation Z bearing the brunt.

In previous waves of layoffs, older employees with higher pay were often the first targets. But the logic in the AI era is reversing.

According to Bloomberg on May 16, the 2026 CEO agenda survey released by Oliver Wyman together with the New York Stock Exchange shows that more than 40% of CEOs plan to reduce entry-level roles within the next one to two years, shifting the workforce structure toward mid- and senior-level employees; only 17% of CEOs say they will increase the proportion of entry-level roles. A year ago, these figures were almost completely the opposite.

The survey covered 415 CEOs (266 from publicly listed companies, 149 from private enterprises). Publicly listed companies account for about 10% of global market capitalization, including 65 CEOs of Fortune 500 companies.

John Romeo, head of the Oliver Wyman Forum, told Bloomberg in an interview, saying, “The difficulty for entry-level employees to enter the workforce is indeed increasing. CEOs now value mid- and senior-level employees more for driving productivity.”

43% of CEOs cut entry-level roles—an AI “seniority bias” effect emerges

The logic behind the shift is not hard to understand: the work that AI agents can handle right now is highly concentrated in the typical task categories performed by entry-level employees. Writing code, assessing sales leads, reviewing documents, organizing data reports—these routine cognitive tasks that used to be done by entry-level staff are being rapidly replaced by AI systems.

But what AI cannot currently replicate is judgment that takes years of industry accumulation to form. Ravin Jesuthasan, a future-of-work consultant, told Bloomberg that companies’ attitude is becoming, “I need someone who has truly done these things—because her experience, judgment, and problem-solving ability make her more valuable than AI.”

This phenomenon already has supporting data in academia. In a paper, Harvard researchers Seyed M. Hosseini and Guy Lichtinger analyzed resume and hiring data covering 62 million employees across 285,000 U.S. companies. The results show that since early 2023, among companies that have actively adopted generative AI, the number of entry-level employees declined by 7.7% relative to non-adopting companies over six quarters, while the number of senior employees was basically unaffected. The key finding is that this drop mainly came from slowing hiring rather than large-scale layoffs. In other words, it’s not that they are cutting people—it’s that they are no longer hiring.

Oliver Wyman’s report puts the implications of this trend more directly: “CEOs with the longest planning cycles are most likely to plan to reduce headcount. This suggests they expect organizations to become leaner under AI augmentation—not as a tool to cut costs, but as an endpoint state.”

Goldman Sachs estimate: AI eliminates about 16,000 U.S. jobs per month on net, with Generation Z hit first

In a research report in April, Goldman Sachs economist Elsie Peng estimated that over the past year, the AI substitution effect eliminated about 25,000 jobs per month, while the AI augmentation effect added about 9,000 jobs per month over the same period—net eliminating about 16,000. Frontierbeat

The impact distribution is highly uneven. In occupations with the highest exposure to AI substitution, the unemployment-rate gap between entry-level employees under 30 and experienced employees aged 31 to 50 has widened dramatically compared with pre-pandemic levels. The pay gap has worsened in parallel. Goldman’s regression analysis estimates that for each one-standard-deviation increase in AI substitution exposure, the salary gap between entry-level and experienced employees widens by about 3.3 percentage points. Fortune

Generation Z is disproportionately concentrated in routine white-collar roles such as data entry, customer service, legal support, and bill processing—exactly the areas where AI is best at automation. They lack the experience “buffer” that senior employees have. Fortune

A study by Stanford University last November further corroborated this: in the fields with the highest AI exposure, the likelihood of unemployment among young employees is 16% higher than for other groups. Fortune

The long-term risk of a broken talent pipeline

Cutting entry-level roles can reduce costs and improve efficiency in the short term, but its hidden costs are prompting concern.

Helen Leis, global head of leadership and transformation at Oliver Wyman, pointed out to Bloomberg that if companies hope to have mid- and senior-level talent in the future to manage AI-driven workflows, “these people need to learn the work inside the company first.” Not hiring entry-level employees is equivalent to cutting off the company’s own talent pipeline.

Andrew McAfee, co-director of the MIT Initiative on the Digital Economy, previously expressed a similar concern to Harvard Business Review: besides learning on the job and apprenticeships, how else can people learn to do a job?

Monster’s survey shows that among the Class of 2026 graduates, nearly 90% are worried that AI or automation will replace entry-level roles, up sharply from 64% in 2025. Fortune

These worries are not unfounded. According to a SignalFire report, between 2023 and 2024, the hiring volume for entry-level positions at 15 of the largest U.S. tech companies declined by 25%. The situation in the U.K. is even more severe: in 2024, tech graduate roles decreased by 46%, and it is expected to fall by an additional 53% by 2026. IEEE SpectrumRezi

A small number of companies go against the trend—AI “winners” place greater emphasis on entry-level employees

Interestingly, the companies that have deployed AI most successfully show a different talent strategy.

Oliver Wyman’s report notes, “A group of advanced AI adopters with counterintuitive strategies believe that this technology increases the value of entry-level talent rather than replacing them.” In companies with higher AI investment returns, the share allocated to entry-level roles is higher than in companies that have not yet seen returns.

In February this year, IBM announced that it plans to expand U.S. entry-level hiring to three times the previous amount and rewrite job descriptions for the AI era. This week, Salesforce CEO Marc Benioff announced the hiring of 1,000 new graduates and interns to build its AI systems, writing on the X platform that, “They say AI will eliminate entry-level jobs, but these graduates and interns are building AI.” Meanwhile, Matt Garman, CEO of Amazon Web Services, publicly stated that replacing entry-level employees with AI is “one of the dumbest decisions a company can make,” because entry-level employees are often the most proficient users of AI tools.

But these cases still make up a minority within the overall trend. Oliver Wyman’s survey shows that 74% of CEOs are freezing or reducing headcount, higher than 67% last year. The most aggressive layoffs occur in the tech, media, and telecommunications sectors.

AI ROI dilemma: most companies are still in the “burning money and trying things out” stage

There is a significant gap between CEOs’ confidence that AI will change workforce structures and the actual returns AI delivers.

Oliver Wyman’s survey shows that 67% of companies’ AI deployments are still in the planning and pilot stages. 53% of CEOs say it is too early to evaluate AI investment returns—up from 41% last year. Only 27% of CEOs say AI investment returns have met or exceeded expectations, down from 38% last year. Nearly a quarter say AI has had no impact on revenue.

The report describes this as “not a confidence crisis, but an awareness of the need for large-scale redesign: this is slower and more difficult than the initial enthusiasm suggested.”

However, the report also points out that for companies deploying AI across more than two application scenarios, their reported cost savings and revenue growth are about twice those of companies deploying in a single scenario. The value curve of AI is nonlinear; real returns concentrate after large-scale deployment.

A single sentence by Teresa Ghilarducci, an economist from the new school, may best summarize the situation today: even if AI tilts the job market toward senior employees, it does not mean they get job security. “Corporate commitments to employees are getting weaker.”