AI human trainers weigh high pay against job risk
NEW YORK, UNITED STATES — In a tightening white-collar job market, the gig economy is emerging, where skilled professionals are hired to train the artificial intelligence (AI) that may one day automate their roles.
In a report by The Wall Street Journal, Bay Area AI startup Mercor, valued at $10 billion, has enlisted over 30,000 contractors in 2025, including experts ranging from poets to lawyers, to critique AI outputs for clients such as OpenAI and Anthropic.
While the work offers high hourly wages, it forces a generation of workers to grapple with the direct consequences of their labor on their professions.
High-paid experts train AI that could replace their jobs
Economic uncertainty and rising unemployment have driven professionals to apply for hundreds of jobs. For many, this now includes high-paying contract work with Mercor, which seeks subject-area experts ranging from dermatologists earning $250 an hour to poets making $150 an hour to refine AI models.
This demand provides a crucial financial lifeline for workers like Peter Valdes-Dapena, a laid-off automotive journalist who now works 20 to 30 hours a week, critiquing AI-generated articles.
However, this labor directly trains AI on human expertise, creating a dark irony for the very workers who fuel it. Contractors like video editor Williams acknowledge joking about training AI to take her job, while co-workers on Slack express unease but feel their job prospects are limited.
The work offers challenging, billable hours monitored by time-tracking software to prevent contractors from using AI to critique AI. Yet, it fundamentally involves optimizing systems that could render similar human tasks obsolete.
As Valdes-Dapena notes, while he didn’t invent AI and cannot uninvent it, his participation underscores a widespread resignation to an inevitable technological shift.
“Many of the people we work with already see AI as inevitable in their field, but that doesn’t mean humans will run out of meaningful work,” said a Mercor spokeswoman in a written statement.
Contractors face IP risks, low leverage in AI data supply chain
The structure of AI training work raises significant ethical and contractual questions, leaving workers with low bargaining power despite their specialized knowledge. Mercor’s hiring process, which includes AI-conducted interviews and sometimes requires screen sharing, leads to contracts that some interpret as overreaching.
Academic Laura Kittel reviewed a contract she believed granted royalty-free rights to her existing and future academic papers for any client’s benefit, a clause she tried unsuccessfully to amend before being told by an AI assistant to accept it or walk away, an experience she described as having her dignity stripped.
Mercor states that its contract applies only to intellectual property used in the course of a project, exempting a worker’s independent creations. Nonetheless, the dynamic illustrates the tension in this nascent industry.
While some contractors, like Sara Kubik, an attorney, find the work rewarding and doubt AI will fully replace complex professions like law, they still confront criticism from peers who label the company a scam or accuse them of AI taking jobs.
While some contractors, like attorney Sara Kubik, find the work rewarding and doubt AI will fully replace complex professions like law, their firsthand experience often reveals a more constrained reality.
Through her contracting, Kubik has seen past the grandiose hype surrounding AI’s potential, saying, “It’s taught me the limitations of AI.” Nevertheless, they still confront criticism from peers who label the company a scam or accuse them of aiding AI in taking jobs.
This underscores a systemic vulnerability, highlighting the urgent need for portable benefits and ethical data-rights frameworks to protect the skilled laborers building the very engines of possible work disruption.
