New research from the Federal Reserve Bank of Dallas reveals a paradox at the heart of AI's labor market impact: in the most AI-exposed industries, employment is falling while wages are rising. The explanation lies in who AI replaces versus who it empowers.
The study by J. Scott Davis, published in February 2026, provides some of the first hard wage data showing that AI is simultaneously aiding experienced workers and displacing entry-level ones - a pattern with significant implications for workforce planning, outsourcing, and virtual assistant services.
The Core Finding: Employment Down, Wages Up
Since ChatGPT launched in fall 2022, the US labor market has shown a striking divergence in AI-exposed sectors:
| Metric | National Average | Computer Systems Design | Top 10% AI-Exposed Industries |
|---|---|---|---|
| Employment change | +2.5% | -5.0% | -1.0% |
| Weekly wage growth | +7.5% | +16.7% | +8.5% |
The numbers tell a clear story. In computer systems design - one of the economy's most AI-exposed industries - employment dropped 5% while wages surged 16.7%. Across the top 10% of AI-exposed industries overall, employment declined 1% while wages grew 8.5%, outpacing the national average.
This is not a contradiction. It is a composition effect: companies are shedding lower-paid, entry-level positions while retaining and paying more for experienced workers whose skills complement AI.
The Experience Premium Effect
The study's most important contribution is quantifying how AI's impact varies by experience level.
In occupations with high experience premiums - where experienced workers earn significantly more than entry-level employees - AI exposure was associated with modest wage gains. In these roles, experience provides something AI cannot replicate: tacit knowledge, judgment, and institutional context.
In occupations with low experience premiums - where entry-level and experienced pay are similar - AI exposure was associated with weaker wage growth. These are roles where the tasks performed are largely codifiable, book-learned, and procedural - exactly the type of work AI automates most effectively.
The practical implication is stark: if your job skills can be learned from a textbook and applied procedurally, AI is coming for your position. If your value comes from years of pattern recognition, relationship building, and domain-specific judgment, AI is likely to make you more productive and more valuable.
Impact on Young Workers
The Dallas Fed's research aligns with Anthropic's separate finding of slowing hiring for workers aged 22-25 in AI-exposed occupations. Gen Z workers face a particularly challenging dynamic:
- Entry-level positions shrinking as AI handles routine tasks that previously served as training grounds
- Experience requirements rising as employers seek workers who can manage AI tools rather than perform the tasks AI now handles
- Career ladders compressing as the middle rungs - the routine work that built experience - disappear
This creates a workforce development paradox: young workers cannot gain the experience that makes them AI-complementary without first having access to the entry-level jobs that AI is eliminating.
What This Means for Outsourcing
The Dallas Fed's findings reshape the outsourcing value proposition:
Routine task outsourcing faces AI competition. Companies that outsource purely procedural, codifiable tasks will increasingly consider AI alternatives. Data entry, basic transcription, and template-based processing are the most vulnerable outsourcing categories.
Experience-intensive outsourcing gains value. Outsourced services that require domain expertise, judgment, and relationship management become more valuable as AI handles the routine layer. The premium for experienced outsourced professionals is rising, not falling.
Hybrid outsourcing models win. The optimal approach combines AI for routine processing with experienced human professionals for judgment-dependent tasks. Outsourcing providers that deliver this hybrid model offer the best of both worlds: AI-level speed and scale for routine work, human expertise for complex work.
Implications for Virtual Assistant Services
The Dallas Fed data maps precisely onto the virtual assistant industry's strategic challenge:
VAs with experience premiums are safe. Virtual assistants who bring years of domain expertise - in healthcare administration, legal support, financial services, or executive assistance - fall into the high-experience-premium category where AI complements rather than replaces.
Generalist entry-level VAs face pressure. Virtual assistants whose primary skills are data entry, basic scheduling, and simple email management are in the low-experience-premium category where AI displacement is most likely.
AI proficiency becomes a survival skill. The VAs who thrive will be those who use AI tools to amplify their expertise - handling more complex tasks, serving more clients, and delivering higher-quality output. AI-proficient VAs are the experienced workers in the Dallas Fed's framework: they use AI, they do not compete against it.
Training and upskilling matter. VA companies that invest in training their teams to work with AI tools, develop domain expertise, and build tacit knowledge will retain and grow their workforce. Those that rely on large teams of generalists performing routine tasks will face the same headwinds the Dallas Fed data reveals.
The Dallas Fed's research makes one thing clear: the AI divide in the labor market is not between "safe" and "unsafe" occupations. It is between workers whose value comes from experience and judgment versus those whose value comes from performing tasks that AI can now handle. hire virtual assistants providers have a choice about which side of that divide they operate on.