The AI Jobs Gap: What CEOs Predict vs. What the Data Actually Shows

For two years, CEOs have warned that artificial intelligence would trigger mass layoffs. Headlines about AI replacing workers have become routine. But the actual data tells a more complicated story—one where the feared job apocalypse has not arrived, yet a quieter, more targeted disruption is already reshaping the labor market for young and entry-level workers.

Here is what the latest research and corporate data reveals about the gap between expectations and reality.

CEO expectations are aggressive—but the data is mixed

CEO surveys consistently show that business leaders expect AI to shrink their workforces. Mercer's 2026 Global Talent Trends survey found that 99% of CEOs expect AI and automation to drive headcount reductions within two years . A separate survey of more than 350 public-company CEOs and investors found that 66% plan to freeze or cut hiring through the rest of 2026 .

Yet actual layoffs explicitly tied to AI are far smaller than these expectations suggest. In 2025, U.S. employers cut roughly 55,000 jobs where AI was cited as the primary reason—out of 1.17 million total cuts . In 2026, AI became the top cited reason for job cuts in multiple months, with companies attributing 101,743 cuts to AI in the first half of the year . But even this acceleration remains modest relative to the total workforce.

Some researchers argue that companies may be using AI as a convenient cover for routine restructuring. Oxford Economics and Fortune found that "firms don't appear to be replacing workers with AI on a significant scale" and that companies may cite AI to justify layoffs that would have happened anyway .

Other surveys paint a far less dire picture. KPMG's 2026 U.S. CEO Outlook Pulse Survey found that fewer than 10% of major U.S. company CEOs plan AI-driven job cuts this year, while more than half actually expect to expand hiring thanks to AI investments .

Entry-level employment is bearing the real brunt

While aggregate displacement has been modest, the picture changes dramatically when you look at younger workers.

Stanford's Digital Economy Lab, analyzing ADP payroll data on millions of U.S. workers through June 2026, found "no evidence of widespread, economy-wide job displacement" from generative AI . But the same study found that employment of workers aged 22–25 in AI-exposed occupations declined by 16% relative to other workers, while experienced workers remained stable .

A U.S. Census Bureau working paper confirms an "immediate, sizable, and persistent decrease" in early-career hires (ages 22–24) in AI-exposed industries after ChatGPT's introduction .

Gartner found that 22% of CHROs report that at least one business leader in their organization has stopped hiring for entry-level roles because AI automates tasks traditionally done by junior employees . The World Economic Forum concurs: a hiring slowdown at entry level is "evident," though AI's precise role remains contested .

This pattern—stable employment for experienced workers, declining opportunities for new entrants—represents a structural shift in how companies build their talent pipelines.

Worker retraining: positive but modest and uneven

Retraining is often proposed as the solution for workers displaced by AI. The evidence suggests it helps, but not as much as many hope.

A major review of 56 randomized U.S. studies, published by Anthropic in August 2026, found that job training programs increase employment by 2–3 percentage points and earnings by roughly $1,000 per year . That is statistically significant but not transformative.

New York Fed research on 1.9 million U.S. workforce training spells found that the average quarterly earnings return for AI-exposed workers is about $1,470, but workers who train for AI-intensive occupations face a 29% earnings penalty compared to peers who pursue more general training . In other words, training for an AI job actually pays less than training for a non-AI job.

The Conference Board reports that most organizations focus on helping employees grow in current roles rather than preparing them for future jobs . A separate study found that 35% of employees have received no AI training at all, and only 18% of those who did felt prepared to work independently .

The gap between intent and execution remains large. While 74% of organizations plan to upskill or retrain employees as AI grows, and 87% report AI skill gaps, the actual effectiveness of these programs lags .

The bottom line

The mass AI-driven job apocalypse CEOs predicted has not materialized. What has emerged instead is a more targeted disruption: young and entry-level workers are losing hiring opportunities, while experienced workers remain largely stable. Retraining helps but delivers modest gains, and corporate investment in genuine future-proofing remains uneven. The real story of AI and jobs in 2026 is not about mass layoffs—it is about the quiet hollowing out of the entry-level pipeline.