Why AI companies pay professionals by the hour to check a chatbot's work |
Radiologists, lawyers and engineers are being recruited to train AI systems. Here is what is driving the demand, and what it means for the people who say yes. |
Tamara Patzer
Nov 11, 2026

Anu Brixey is a radiologist. Some weekends, after a full week of reading scans, she opens her laptop and marks where calcium shows up in images of the heart, scan after scan, so an AI system can learn what to look for. Bloomberg reporter Morgan Meaker described her routine in an Oct. 16, 2026, story on the market for expert data work. The story noted that Brixey earns in an hour roughly what another data worker it profiled, a stay-at-home mother who takes short online tasks, earns in a week.
That gap explains a lot about the listings in this newsletter.
Why the demand exists
Early AI systems learned mostly from text and images gathered from the public internet. Built In, a technology news site, reported Nov. 12, 2025, that developers have hit what industry insiders call a "data wall," the point where most of the high-quality material freely available online has already been used. To keep improving, companies need fresh, specific knowledge that cannot simply be scraped, such as a doctor's judgment about a scan or a lawyer's read of a contract. Bloomberg reported that AI companies are now racing to secure private, specialized data to set their next models apart from rivals.
What the work looks like
The job is usually less about inventing and more about reviewing. Built In writer Jeff Rumage described experts spotting a chatbot's inaccuracies, correcting mistakes and applying industry knowledge to a particular situation. Bloomberg cited advertised pay of up to $120 an hour for software engineers, $170 an hour for condensed matter experts and $400 an hour for civil rights attorneys. It also reported that LinkedIn named data annotator the fourth-fastest-growing job role in the United States this year.
A smaller slice than the headlines suggest
Not everyone thinks the expert boom is as large as it sounds. Antonio Casilli, a professor at the Polytechnic Institute of Paris who studies data work, told Bloomberg that highly paid experts take part in relatively few projects compared with lower-paid generalists. He called them "just the cherry on top." Mercor, one of the biggest expert platforms, said its average pay is now $110 an hour but told Bloomberg it did not have data on what share of tasks pay that rate. Prolific, a British platform, said experts make up 30% of its participants but declined to say how many tasks they complete.
Questions the workers raise
Danielle Li, an MIT professor who studies AI and the labor market, told Bloomberg that employers normally rent labor, and that changes when an employer can keep a version of that labor inside a model. Pauley Chea, a radiologist who works as a data worker on the side, said medical students and residents are especially uneasy. Built In made a similar point from the other direction: the work can be monotonous, and opportunities can dry up between projects.
What to take from it
The listings here reflect a real shift in how AI is built. They are also contract work, the pay figures are advertised ranges rather than promises, and projects come and go. Read each full listing before you apply, check the pay terms and decide what the work is worth to you.
I may be paid if you apply through the links on this page. I am not an employee or representative of the company that posts these roles.
Sources: Bloomberg Law, Morgan Meaker, Oct. 16, 2026; Built In, Jeff Rumage, Nov. 12, 2025. |
0 Comments
Join the conversation
Be the first to comment
Share your thoughts above.