Dallas Fed Research: AI Is Creating and Destroying Jobs at the Same Time

Is artificial intelligence coming for your job or making it better? According to new research from the Dallas Federal Reserve, the answer is both — and which side of that equation you land on depends heavily on where you are in your career.

In a February 2026 analysis, Dallas Fed economist J. Scott Davis found that AI is simultaneously reducing employment in the most AI-exposed industries while pushing wages higher for experienced workers in those same fields. The findings paint a nuanced picture that challenges the simple narrative of robots taking all the jobs. For the millions of Americans navigating today's labor market, the research offers a clear signal: experience and tacit knowledge are becoming more valuable than ever, while textbook skills alone are losing ground fast.

What Did the Dallas Fed Actually Find?

The Dallas Fed research examined employment and wage trends across more than 200 occupations since the release of ChatGPT in fall 2022. The results tell two very different stories depending on where you look.

On the employment side, the picture is sobering. Total U.S. employment grew approximately 2.5% since late 2022, but the sectors most exposed to AI tell a different story. Employment in computer systems design and related services has declined 5%. More broadly, employment has dropped 1% across the top 10% of AI-exposed sectors, using an AI exposure index developed by researchers Edward W. Felten, Manav Raj, and Robert Seamans.

But here is where it gets interesting. Despite falling employment, wages in these same AI-exposed sectors are not declining. In fact, they are rising faster than the national average. Since fall 2022, nominal average weekly wages nationwide increased 7.5%, while the computer systems design sector saw wages climb 16.7%. Among the top 10% of AI-exposed industries overall, wages grew 8.5%.

This creates a paradox that demands explanation. If AI were simply automating jobs out of existence, both employment and wages should be falling. The fact that wages are rising even as headcounts shrink suggests something more complex is happening.

Why Are Entry-Level Workers Getting Hit the Hardest?

The Dallas Fed research draws on a critical distinction that helps explain what is unfolding: the difference between codified knowledge and tacit knowledge.

Codified knowledge is the kind of information you can learn from a textbook, an online course, or a training manual. It is structured, documented, and transferable. Tacit knowledge, on the other hand, is the understanding that comes only through years of hands-on experience — the instincts, judgment calls, and pattern recognition that cannot easily be written down or taught in a classroom.

AI, as it turns out, is very good at replicating codified knowledge but struggles with tacit knowledge. This has profound implications for who wins and who loses in the AI-driven labor market.

For entry-level employees fresh out of school, the expert part of their job is the codified knowledge they spent years studying. When AI can perform those same tasks faster and cheaper, the value proposition of hiring a new graduate weakens considerably. Research from Stanford University by Erik Brynjolfsson, Bharat Chandar, and Ruya Chen confirms this pattern, showing that the recent employment decline in AI-exposed sectors is particularly pronounced for workers under age 25.

Dallas Fed economist Tyler Atkinson adds an important detail: this decline is not primarily driven by layoffs. Instead, young workers entering the labor force are facing dramatically lower job-finding rates. The jobs simply are not there for new graduates in AI-exposed fields at the same rate they used to be.

For experienced workers, the dynamic flips. The codified tasks that AI can handle represent the routine, less valuable parts of their work. By automating those tasks, AI frees up experienced professionals to focus on the higher-value work that requires their tacit knowledge — the judgment, creativity, and strategic thinking that comes from years in the field.

How Does Experience Change the AI Wage Equation?

To quantify this dynamic, Davis examined Bureau of Labor Statistics wage estimates for entry-level versus experienced workers across more than 200 occupations. He calculated an "experience premium" for each occupation — the percentage difference between what experienced workers earn and what entry-level workers earn.

The results are striking. The median experience premium across all occupations is 40%. But the range is enormous. Fast-food cooks, ticket agents, and dry cleaners have experience premiums below 10%, meaning there is little wage difference between new and veteran workers. Lawyers, insurance underwriters, credit analysts, and marketing specialists have experience premiums exceeding 100%, meaning experienced workers earn more than double what entry-level workers make.

When Davis combined these experience premiums with AI exposure data, a clear pattern emerged. For occupations with a 0% experience premium — where experience does not matter much — increased AI exposure is associated with a 0.28 percentage point reduction in wage growth. AI is pushing wages down because it can substitute for both new and experienced workers equally.

For occupations at the 90th percentile of experience premiums, the story reverses. Increased AI exposure is associated with a 0.2 percentage point increase in wage growth. In these fields, AI is complementing experienced workers and making their expertise more valuable, not less.

The takeaway is clear: if your job rewards experience and tacit knowledge, AI is likely working in your favor. If your role relies primarily on codified, textbook skills, you are in a more vulnerable position regardless of your seniority.

What Does This Mean for the Future of Career Development?

The Dallas Fed findings raise a fundamental challenge for how companies train and develop talent. The traditional white-collar career path involves taking an entry-level position right out of school, performing codifiable tasks, and gradually building the tacit knowledge needed to become a senior professional. That model is now under pressure.

If AI can handle the codifiable work that entry-level employees used to do, companies face a dilemma. In the short term, it is cheaper to let AI do that work. But in the long term, someone needs to develop the tacit knowledge that makes experienced workers so valuable. Without entry-level roles serving as training grounds, the pipeline of future experienced talent dries up.

As Davis writes, "leaving new employees off the job ladder is not sustainable in the long run. In the long run, AI adoption will require rethinking how entry-level employees gain experience on the job."

This is not just a corporate strategy question. It is a societal one. If the traditional path from entry-level worker to experienced professional breaks down, entire industries could face talent shortages a decade from now.

As Metaintro CEO Lacey Kaelani told ZDNet, "A large number of our best product ideas have come from engineers doing the same repetitive data validation work over and over again, where they notice patterns that would lead to larger insight. However, once we eliminated that repetitive task and automated it, we definitely improved automation, but we lost the incidental learning that happens [through] seeing the data."

Kaelani's observation highlights a hidden cost of AI automation that the Dallas Fed data supports at a macro level. When companies eliminate entry-level tasks, they do not just cut costs. They also cut off the experiential learning pipeline that produces the experienced workers AI cannot replace.

Which Industries and Roles Are Most Affected?

The Dallas Fed data, combined with the AI exposure index from Felten, Raj, and Seamans, points to several sectors where the dual impact is most visible.

Technology and computer systems design has been the hardest hit on the employment side, with a 5% decline in headcount since late 2022. Yet wages in the sector have surged 16.7%, suggesting that companies are shedding junior roles while paying more for senior talent whose experience AI cannot replicate.

Financial services occupations like credit analysts and insurance underwriters carry high experience premiums (over 100%), making them prime candidates for AI augmentation rather than replacement. Experienced professionals in these fields are seeing wage benefits from AI adoption.

Legal services similarly reward tacit knowledge heavily. Junior associates performing document review and legal research face pressure from AI tools, while senior attorneys whose value lies in courtroom strategy, client relationships, and judgment calls are likely to see their expertise appreciated even more.

Marketing and creative roles show a similar split. AI can generate basic copy, analyze data, and automate campaign management — tasks that fall on entry-level marketers. But the strategic thinking, brand intuition, and creative direction that come from experience remain firmly in human hands.

On the more vulnerable end, occupations with low experience premiums — where the gap between entry-level and experienced wages is small — face the greatest risk. These include roles where tasks are highly codifiable and experience does not add much differentiated value.

How Should Job Seekers and Workers Respond?

The Dallas Fed research carries practical implications for anyone navigating today's labor market.

For new graduates and early-career workers, the message is urgent. Codified skills alone are no longer a reliable ticket to employment. Seek out roles, internships, and projects that build tacit knowledge as quickly as possible. Look for positions where you work alongside experienced professionals and learn the judgment calls, pattern recognition, and contextual understanding that AI cannot replicate. Prioritize depth of experience over breadth of textbook credentials.

For mid-career professionals, the news is more encouraging. Your years of experience are becoming more valuable, not less. The tacit knowledge you have built — understanding how your industry really works, recognizing patterns that no textbook captures, making judgment calls under uncertainty — is exactly what AI cannot do. Lean into that advantage by positioning yourself as someone who uses AI tools to amplify your expertise rather than being replaced by them.

For career changers, consider targeting occupations with high experience premiums. Fields like law, finance, marketing strategy, and specialized consulting reward experience heavily, meaning AI is more likely to complement your work than compete with it.

For employers, the research is a wake-up call about talent development. If you let AI replace all entry-level work without creating alternative pathways for developing tacit knowledge, you will face a shortage of experienced talent in five to ten years. Invest now in apprenticeship models, mentoring programs, and hybrid roles that let new workers build experience even as AI handles routine tasks.

People Also Asked

Q: Will AI replace entry-level jobs completely?

A: Not completely, but the Dallas Fed data shows entry-level workers in AI-exposed sectors are facing significantly tougher job markets. Employment for workers under 25 in these fields has declined notably since late 2022, primarily because companies are hiring fewer new graduates rather than laying off existing workers. Entry-level roles that rely heavily on codified, textbook knowledge are most at risk. However, positions that involve hands-on learning, client interaction, and building tacit knowledge will continue to exist because companies need to develop their future experienced talent.

Q: What is the difference between codified knowledge and tacit knowledge?

A: Codified knowledge is information that can be written down, taught in a classroom, or documented in a manual — think formulas, procedures, and standard practices. Tacit knowledge is understanding that comes only through experience, like recognizing subtle patterns in data, reading a room during a negotiation, or knowing which approach will work in a specific context. AI excels at replicating codified knowledge but cannot yet replicate tacit knowledge, which is why experienced workers with deep domain expertise are seeing wage gains rather than job losses.

Q: Which jobs are safest from AI replacement?

A: According to the Dallas Fed research, jobs with the highest "experience premiums" — where experienced workers earn significantly more than entry-level workers — are the safest. These include lawyers, insurance underwriters, credit analysts, and marketing specialists, all of which have experience premiums exceeding 100%. In these occupations, AI exposure is actually associated with positive wage growth because the technology complements rather than replaces the tacit knowledge that experienced professionals bring. Jobs with low experience premiums, where new and veteran workers earn similar wages, face the greatest risk.

Future-proof your career by staying ahead of how AI is reshaping the job market. Metaintro delivers the latest workforce intelligence, career strategies, and hiring trends straight to your inbox so you can make smarter moves in an AI-driven economy. Sign up today and take control of your career trajectory.