The most striking finding is a 16% decline in employment among workers aged 22 to 25 in sectors particularly vulnerable to AI, such as customer service and software development.

AI Is Eliminating Jobs for Younger Workers – A Stark Warning or a Complex Shift?

by ChatGPT-4o

A new WIRED article by Will Knight, based on research from Stanford University, offers one of the clearest signals yet that artificial intelligence—particularly generative AI—is not merely reshaping the workforce in abstract terms, but is already eliminating jobs, with younger workers facing the brunt of the displacement. The research, spearheaded by Erik Brynjolfsson and colleagues Ruyu Chen and Bharat Chandar, used payroll data from ADP (the largest payroll provider in the U.S.) covering the period from the debut of ChatGPT in late 2022 through mid-2025. The results are troubling but nuanced, providing both a warning and a blueprint for adaptation.

Evidence of a Shift

The most striking finding is a 16% decline in employment among workers aged 22 to 25 in sectors particularly vulnerable to AI, such as customer service and software development. These sectors are often characterized by repetitive, lower-experience roles—the types most easily automated by large language models and generative AI tools. This finding cuts through the ambiguity that has long surrounded automation debates, shifting the discourse from hypothetical job loss to measurable displacement.

What’s more, this trend persisted even when adjusting for potential confounding factors like post-pandemic labor disruptions, remote work trends, and waves of tech layoffs. The conclusion is clear: AI’s footprint in the labor market is distinct and tangible, particularly for younger, less experienced workers.

Experience as a Buffer

In a perhaps surprising twist, the study also revealed that more experienced workers within the same AI-affected industries were largely insulated from job loss. Some even saw a slight increase in opportunities. This suggests that AI is currently displacing “low-value” tasks rather than entire professions—at least for now. For instance, routine coding tasks or API connections are easily handled by AI, while the oversight, design, and integration of these outputs still require human judgment.

Thus, AI is not evenly distributed in its disruption. It is hitting younger workers hardest—those who typically enter the workforce through entry-level roles where repetition is common and training is minimal. Ironically, these are also the jobs that traditionally serve as stepping stones to higher-skill careers.

No Wage Collapse—Yet

An interesting counterpoint in the study is that while jobs have declined in some areas, wages have not followed the same trend—at least not yet. This supports the idea that AI is not driving down labor value uniformly but is instead shifting demand toward roles that cannot easily be automated. This includes human oversight roles, hybrid jobs that require AI fluency, and positions emphasizing soft skills and strategic thinking.

Policy and Design Recommendations

Erik Brynjolfsson and Andrew Haupt advocate for designing AI systems that enhance human capabilities rather than replace them—a model sometimes referred to as "centaur AI." This approach calls for AI benchmarks that evaluate systems based on their ability to augment human performance. It’s a response to the growing recognition that many future roles will require a partnership between human workers and AI, rather than a binary competition between the two.

Another policy recommendation includes rethinking tax incentives that reward automation over labor. If businesses are incentivized to replace workers with machines without accounting for long-term social costs, the displacement trend may accelerate dangerously. Public policy needs to catch up fast to ensure innovation does not come at the cost of generational employment and societal cohesion.

Implications for the Future

Matt Beane of UC Santa Barbara suggests that we are on the cusp of a “mountain of augmentable work” where AI’s output will need constant steering, verifying, and contextualizing by humans. This opens up a new job class—one that is AI-literate but not necessarily technical. However, access to these jobs may remain limited to those who already have a head start: educational background, mentorship, and existing experience.

Brynjolfsson’s call for an “early-warning system” to track real-time displacement is especially timely. Given the pace of AI advancement, what starts as a problem for younger workers could rapidly extend to mid-career professionals if safeguards and proactive workforce planning are not put in place.

Conclusion: What Must Be Done

The article paints a complex but urgent picture. On one hand, AI is creating new possibilities and efficiencies; on the other, it is already harming the youngest entrants into the job market—those with the least power and fewest safety nets. If unaddressed, this dynamic could have long-term consequences for economic mobility, intergenerational equity, and social stability.

To respond effectively, stakeholders should consider the following actions:

• Governments must develop real-time labor tracking tools, reform tax codes that currently reward labor-replacing automation, and invest in AI-human collaboration models.

• Businesses should prioritize AI strategies that augment rather than eliminate human work, and ensure young workers are trained in AI fluency as part of onboarding and development.

• Educators and career services must rethink entry-level pathways and design curricula that reflect the new reality—where routine work is automated but human judgment remains indispensable.

• Young workers themselves should invest in hybrid skills—combining domain expertise with AI literacy—to future-proof their careers.

AI’s march is not a distant threat—it is already reshaping the lower rungs of the job ladder. The task now is to ensure that the ladder itself remains climbable.