Employment data shows early-career roles are shrinking in AI-exposed sectors, while risks for older workers loom.

Artificial intelligence has already shifted from the realm of theory to the fabric of daily work. For years, debates have swirled around whether AI would primarily create jobs, eliminate them, or simply change their nature. But for those entering the workforce today, the impact is no longer hypothetical; it is measurable. A new paper from Stanford University’s Digital Economy Lab, authored by Erik Brynjolfsson, Ruyu Chen, and Bharat Chandar, provides striking evidence of what many suspected: AI is displacing young professionals at the very start of their careers.

Using data from payroll processing firm ADP, which tracks more than 25 million U.S. workers, the researchers found that workers aged 22 to 25 in the most AI-exposed jobs experienced a 13 percent relative decline in employment between late 2022 and mid-2025. While some industries, particularly healthcare and services requiring direct human contact, saw employment for younger workers rise, entry-level roles in highly automated fields are evaporating. This finding sets a dangerous precedent: if AI hollows out the foundation of early-career pathways, the entire pipeline of workforce development may be destabilized.

The six effects of AI on employment trends

Economists generally agree that AI affects employment in six interrelated ways. The Stanford study illustrates these effects vividly:

1. Job Displacement – AI substitutes for certain tasks or roles, leading to outright job losses. The 13% decline for early-career workers in AI-exposed roles is a clear case of this.
2. Job Augmentation – AI complements human workers, increasing efficiency and reducing repetitive tasks. For example, lawyers or consultants using AI tools to summarize documents still keep their jobs but can focus on higher-value work.
3. Job Creation – New industries and positions emerge around AI, from prompt engineers to AI auditors. However, these roles often demand advanced technical skills, leaving younger or less-experienced workers at a disadvantage.
4. Productivity Gains – AI raises efficiency, allowing firms to do more with fewer workers. While this can boost profits, it often translates into slower hiring, especially for entry-level positions.
5. Skill Shifts – The skills in demand evolve rapidly, requiring constant reskilling. Older workers may have institutional knowledge, but younger workers often face the steepest uphill battle to gain entry as the “baseline” job requirements rise.
6. Inequality Risks – AI adoption may widen divides between those who can adapt (often older, more established professionals with resources) and those who cannot (new graduates without experience or connections).

Taken together, these effects show why the burden of AI’s impact falls most heavily on the young. They are the first to lose entry-level footholds, the last to benefit from augmentation, and the least likely to access newly created jobs without prior experience.

A dangerous precedent for the future

Early-career positions are more than just income sources; they are launch pads. They teach professional norms, communication skills, and industry-specific practices that no classroom can fully replicate. If AI wipes out large portions of these roles, an entire generation risks being deprived of the opportunity to “learn by doing.”

This is not just a youth problem. While older and more experienced workers appear safer today, protected by their accumulated skills and institutional value, they are not immune. As AI systems grow more advanced, even higher-level decision-making, management, and creative work will face substitution. What begins with young graduates today may foreshadow what happens to seasoned professionals tomorrow.

In other words: AI is moving up the ladder.

Augmentation vs. substitution: a corporate choice

The Stanford study also highlights a critical distinction: whether firms adopt AI to replace workers or to augment them. In industries where AI assists rather than substitutes, job losses are smaller, and workers can focus on higher-value tasks. For example:

• In healthcare, AI can analyze patient data, leaving doctors to prioritize care.
• In education, AI can automate grading, freeing teachers for mentorship.
• In law, AI can perform initial case research, giving junior associates more time to hone strategy.

But in industries where AI directly replaces roles, from call centers to back-office functions, opportunities for young workers vanish, creating a ripple effect that will haunt future labor supply chains.

What’s at stake?

If current trends continue, the short-term savings businesses gain from automation could be outweighed by long-term risks. Without a healthy flow of young professionals entering and training in the workforce, organizations will struggle to sustain innovation, leadership, and growth. Moreover, societies will face widening generational divides, with the youth excluded from economic opportunity while older workers cling to temporary security.

Policymakers and educators must respond by strengthening pathways into AI-resilient roles and focusing training on the kinds of problem-solving, creativity, and human judgment that machines cannot easily replicate. But corporations, too, bear responsibility. Leaders must ask: Are we building a sustainable workforce by using AI to empower human capital, or are we eroding our own future by hollowing out the base of the talent pyramid?

A Choice, Not Destiny

We stand at a crucial juncture. The decisions made today about AI deployment will shape the economic opportunities available to an entire generation. If we choose automation over augmentation, we risk creating a future where AI's productivity gains are concentrated among those already established in their careers, while newcomers are systematically excluded from entire industries.

Alternatively, if we choose to implement AI as a tool that amplifies human capability rather than replaces it, we can create a future where technology serves as a bridge to opportunity rather than a barrier to entry. The data from Stanford provides both a warning and a roadmap. The question is whether we will heed it.

 Originally Published on LinkedIn.