Freelancers Hit Hard by AI — And the Highest-Skilled Are Losing the Most

If you're a freelancer who writes, designs, or edits for a living, you've probably felt the ground shifting beneath your feet since late 2022. A new study published through the Brookings Institution puts hard numbers on what many have suspected: generative AI is not just changing how freelance work gets done — it's shrinking the market for the people who do it.

The research, conducted by economists Xiang Hui and Oren Reshef from Washington University in St. Louis, analyzed millions of contracts on a major online freelancing platform. Their findings are sobering, and one conclusion in particular turns conventional wisdom about technology and skills completely on its head.

The Numbers Tell a Clear Story

After the release of tools like DALL-E 2, Midjourney, and ChatGPT between mid-2022 and late 2022, freelancers in AI-exposed occupations saw their monthly contracts drop by roughly 2% and their total monthly earnings fall by about 5% [Fact]. That gap between contracts and earnings matters — it means clients aren't just hiring fewer freelancers, they're also paying less for the work that does get commissioned.

What makes this particularly concerning is the trajectory. The impact didn't spike and stabilize. Instead, the decline built steadily over six to eight months [Fact], suggesting this isn't a temporary shock that the market is absorbing. It looks more like a structural shift.

The occupations hit hardest were precisely the ones you'd expect: copyediting, proofreading, and graphic design [Fact]. These are tasks where generative AI can produce a "good enough" first draft — or even a final product — at a fraction of the cost. Meanwhile, occupations like administrative services, video editing, and data entry showed much less impact [Fact], likely because they involve coordination, judgment calls, or physical-world context that current AI handles poorly.

The Surprise: Experience Became a Liability

Here's where the study delivers its most counterintuitive punch. In previous waves of technological disruption — think ATMs replacing bank tellers, or spreadsheets reshaping accounting — it was typically the lower-skilled, less experienced workers who bore the brunt. Senior professionals had accumulated enough expertise and client relationships to weather the storm.

With generative AI, the pattern is reversed [Fact].

Highly skilled, experienced freelancers in AI-exposed fields lost more work than their junior counterparts. The researchers suggest this is because AI effectively compresses the skill distribution. When a client can use ChatGPT to produce polished copy or Midjourney to generate professional-looking graphics, the premium that used to justify hiring a seasoned expert evaporates. A junior freelancer charging an hour was already competing on price; a senior one charging an hour was selling quality and reliability — exactly the gap that AI narrows.

This is a fundamentally different dynamic from what labor economists have observed in decades of automation research [Claim]. If it holds across other sectors — and early signs suggest it might — it means the traditional advice of "upskill to stay safe" needs serious re-examination.

What This Means for Your Career

The Brookings study focuses on freelancers, but the implications ripple outward. According to McKinsey's latest survey, 71% of companies now use generative AI regularly in at least one business function [Fact]. That adoption rate was just 33% in early 2023. The freelance market, with its rapid contract cycles and price transparency, is simply where the effects show up first.

If you're a graphic designer, the data suggests that pure execution work — creating layouts, generating visual assets, basic brand materials — is where AI competition is fiercest. Our analysis shows graphic designers already face an AI automation exposure of 72/100 on task-based measures. Designers who thrive will likely be those who move toward creative direction, brand strategy, and the kind of contextual judgment that requires understanding a client's business deeply. See detailed data for graphic designers

For editors and proofreaders, the picture is similarly challenging. AI can now catch grammar errors, suggest rewrites, and even match a publication's style guide with reasonable accuracy. But it still struggles with narrative structure, voice consistency across long documents, and the kind of editorial judgment that makes a piece genuinely compelling rather than merely correct. Explore the full breakdown for editors

Web developers face a more nuanced situation. While AI code generation tools have exploded in popularity, the freelance market data shows less dramatic impact here than in writing and design [Fact]. This may be because software projects involve more back-and-forth iteration, debugging, and integration work that clients can't easily hand off to an AI tool alone. Still, developers focused purely on templated websites or basic front-end work should take note. Check AI exposure data for web developers

For writers and authors, the challenge is existential but also opportunity-rich. Content mills and commodity writing are clearly in AI's crosshairs. But long-form journalism, personal essays, specialist technical writing, and narrative nonfiction require the kind of research depth, lived experience, and stylistic voice that AI cannot replicate convincingly. View the analysis for writers and authors

Practical Steps Forward

The Brookings research doesn't paint a hopeless picture — it paints an urgent one. Here's what the data suggests you should consider:

Redefine your value proposition. If your freelance pitch is "I produce X quickly and affordably," you're competing directly with AI. Shift toward "I understand your business well enough to produce X that actually achieves Y." The strategic layer is where human value persists.

Learn to work with AI, not against it. Freelancers who integrate AI tools into their workflow can potentially deliver more value in less time. The risk is in resisting the tools entirely or in letting them commoditize your output.

Diversify your skill set. The study shows that narrow specialists in AI-exposed tasks are most vulnerable. Cross-disciplinary skills — a designer who also understands data visualization, a writer who can also do UX research — create combinations that AI can't easily replicate.

Watch the trajectory, not just the snapshot. That six-to-eight-month deepening trend is perhaps the most important data point in the entire study. The impact of generative AI on freelance markets is not stabilizing. It's accelerating [Claim].

The freelance market has always been a canary in the coal mine for broader labor trends. What's happening there now is likely a preview of changes coming to traditional employment in the months and years ahead.

Sources

• Hui, X. & Reshef, O. (2025). "Is generative AI a job killer? Evidence from the freelance market." Brookings Institution
• McKinsey Global Institute. (2024). "The state of AI in early 2024." McKinsey & Company

Update History

• 2026-03-21: Initial publication based on Brookings Institution research by Hui & Reshef (2025).

---

This article was generated with AI assistance using data from the cited sources. All factual claims are attributed and tagged with confidence indicators. For detailed occupation-level data, visit the individual occupation pages linked above. Learn more about our AI-assisted content process.