The AI training gig used to mean tagging photos of stop signs for pocket change. That job still exists, but it now has a far more expensive cousin.
Gig workers holding medical, law and engineering degrees are now training AI models and billing hundreds of dollars an hour for it, according to a Bloomberg report published October 16, 2026. The report describes AI training shifting away from low-wage, generalist data labeling and toward specialists in fields like medicine, law and engineering. These experts check AI outputs for accuracy, give detailed feedback, and evaluate complex reasoning within their own disciplines.
The going rate for a professional opinion
Experts with advanced degrees are earning between $80 and $450 per hour on specialized AI tasks as of October 2026, depending on their qualifications. For comparison, generalist data workers earn between $10 and $65 per hour. At the very top of the market, the best specialists can earn up to $1,000 per hour.
Mercor is one of the platforms leading this trend. The company has approximately 30,000 active contractors and pays out between $1.5 million and $4 million daily to those contractors. The average contracted rate for experts on Mercor sits around $85 to $108 per hour, with workers holding advanced credentials commanding higher rates.
As of mid-October 2026, thousands of AI training roles were open, with the heaviest concentration in law, medicine and finance.
Why AI suddenly needs a law degree
Early AI systems needed humans to answer simple questions: is this a cat, is this sentence positive or negative, does this box contain a pedestrian. Today’s models are attempting tasks that look a lot more like professional judgment. Drafting a contract clause, reasoning through a diagnosis or checking an engineering calculation all require someone who knows what a correct answer actually looks like. A generalist labeler can tell you whether a chatbot response reads smoothly. Spotting whether that response quietly misstates a legal standard or a dosage takes a professional.
Bloomberg’s reporting ties the strongest demand to professionals with advanced degrees, reflecting AI models that have matured enough to require deeper domain knowledge.
What this means for workers, platforms and the AI industry
For professionals, this opens an unusual side door. A doctor, attorney or engineer can sell expertise by the hour without a traditional employer, a practice or a billable-hours partnership track. Rates in the $80 to $450 range compete with plenty of conventional professional work.
For AI developers, the economics cut both ways. Paying experts by the hour makes model improvement more expensive, and the costs scale with how specialized the target field is. A legal or medical model that needs constant expert review carries a labor bill that a general chatbot did not.
Platforms like Mercor sit in the middle of this exchange, matching credentialed workers with AI labs that need them. Daily payouts in the millions of dollars suggest the marketplace model has real volume behind it.
Watch also the gap between generalist and specialist pay. If the $10-to-$65 tier and the $80-to-$450 tier keep pulling apart, AI training could settle into a two-class labor market: commodity annotators on one side, credentialed consultants on the other.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.