30-06-2025
Exclusive: AI job substitution is real, but narrow, per new study
Generative AI is replacing low-complexity, repetitive work, while also fueling demand for AI-related jobs, according to new data from freelance marketplace Upwork, shared first with Axios.
Why it matters: There are plenty of warnings about AI erasing jobs, but this evidence shows that many workers right now are using generative AI to increase their chances of getting work and to boost their salary.
The big picture: Uncertainty around AI's impact and abilities means companies are hesitant to hire full-time knowledge workers.
Upwork says its platform data offers early indicators of future in-demand skills for both freelancers and full-time employees.
By the numbers: Freelance earnings from AI jobs are up 25% year-over-year, per the report.
Freelancers in AI earn over 40% more per hour than those doing non-AI work.
What they're saying: The increased freelance earnings from AI jobs are typically from people who already had experience in that particular field, Kelly Monahan, manager of the Upwork Research Institute, tells Axios.
"If you were a traditional machine learning expert, and now you're augmenting that work with generative AI, you're seeing such a great premium," Monahan says.
The same is true for graphic designers who use AI image or video generation tools, she adds.
Between the lines: Most business leaders still don't trust AI to automate tasks without a human in the loop, so they're keen on anyone who knows how to use AI to augment their work.
Upwork data also showed that managers still trust humans working alone and humans working with AI over AI-only outputs. And that trust gap is widening.
Workers using AI for augmentation outnumber those using it for automation by more than 2 to 1, a trend that matches data that Anthropic released in February.
Demand for repetitive coding has dipped, but the report shows clients are still looking for experienced developers for more complex projects.
Freelancers with jobs requiring coding skills for at least 25% of the work now earn 11% more for the same jobs compared to November 2022, when ChatGPT launched, per Upwork.
The data also suggests that the rise of vibe coding — using AI to code without knowing how to code — is creating demand for more workers who are skilled at using vibe coding tools.
This is giving rise to what Upwork calls "the generalist," which is anyone who can work with AI to code and design.
What they did: Upwork used its proprietary platform data to evaluate more than 130 categories of work and 62 narrow job categories within broader job fields like "design & creative" or "sales & marketing."
Researchers tracked the full lifecycle of tasks and jobs with the Upwork marketplace over the last six months.
The company says its dataset includes "millions of job posts" and "billions of dollars in freelance earnings," tracking how job types changed based on when contracts started and how much a freelancer earned.
Yes, but: AI adoption in the workplace has had a rocky start with executives pushing for it and workers pushing back.
Many employees still don't understand company policy and are using their own AI tools in secret.
A Duke University study published last month found that workers who use genAI "face negative judgments about their competence and motivation from others."
Zoom out: Anthropic on Friday launched a program to study how its tools and others are changing the way we work.
The Economic Futures Program will provide research grants, work with policy makers and track AI's economic impact and usage trends over time.
"There's no shortage of opinions about AI's economic outcomes," Anthropic wrote in its announcement. "We believe it's fundamental to ground these conversations in real-world data."
Anthropic CEO and co-founder Dario Amodei told Axios last month that AI could wipe out half of all entry-level white-collar jobs and spike unemployment to 10-20% in the next one to five years.
