The AI Workforce Paradox: Trainers vs. Displacement

While a booming industry of freelancers is being hired to teach artificial intelligence the nuances of their professions, economists and industry leaders are sharply divided on whether this trend signals collaboration or impending workforce collapse.

Robbie Hiser, a music producer, is part of Mercor’s expanding network of over 100,000 specialists recruited to refine AI models. Despite helping build systems that could theoretically replicate creative tasks, Hiser remains confident that human experience provides an irreplaceable edge. “AI doesn’t have any life experience to pull from,” Hiser explained, noting that algorithms lack the personal history behind emotions like heartbreak, which are essential for authentic songwriting.

However, his optimism contrasts with a Pew Research study revealing that nearly 75% of Americans fear AI will erode job security. Furthermore, a September report from the McKinsey Global Institute projected that up to 11 million U.S. workers could be forced into career shifts by 2035 as automation reshapes the labor market.

Mercor CEO Brendan Foody views the current shift as a positive evolutionary step. He argues that by automating mundane tasks, AI will liberate professionals to focus on more meaningful work. This philosophy is already being applied in Columbus, Ohio, where the law firm Vorys, Sater, Seymour and Pease partnered with Stanford University’s Liftlab to create AI personas modeled after its senior attorneys.

Kim Herlihy, a trial lawyer and partner at the firm, stated that while the AI can provide feedback based on the firm’s accumulated wisdom, it cannot replace the fundamental human connection required in legal representation. “AI cannot take the place of what’s a fundamentally human business,” she said. “You call your lawyer because you want to look in their eye and say, ‘What should I do about this issue?'”

Despite these assurances, the economic data presents a more complex picture. AI training has become the fourth-fastest-growing job category on LinkedIn, with Mercor gig work paying between $20 and $200 an hour. Yet, Nobel Prize-winning economist Daron Acemoglu warns that these roles are insufficient to counteract broader automation trends.

“The number of people they’re employing in these positions is small relative to the number of people who will be replaced,” Acemoglu said. He highlighted that the current AI revolution is occurring with unprecedented speed, compressing into one or two years what took the Industrial Revolution 80 years to achieve. In a worst-case scenario, he cautioned that unemployment could triple within the next decade if no policy interventions are made.

Evidence of this disruption is already visible among younger workers. Stanford research indicates hiring has declined for young people in AI-exposed fields such as software development. Similarly, U.S. Census data shows that hiring and wages for recent college graduates in AI-vulnerable majors have dropped.

Clara Shih, former head of AI divisions at Salesforce and Meta, witnessed this efficiency gain firsthand. She noted that tasks previously requiring dozens of employees to conceive, prototype, and deliver products can now be executed by a small team using AI agents. Alarmed by the impact on early-career professionals, Shih resigned in January to launch a nonprofit dedicated to helping young workers navigate the shifting economy.

“It’s never just about the jobs. It’s people’s sense of self-worth. It’s the fracturing of communities,” Shih said, drawing parallels to the decline of manufacturing jobs she observed when her family moved from Hong Kong to Ohio.

Acemoglu emphasized that the solution is not to reject technology, but to steer it. He suggested that companies should design AI to augment human capabilities rather than replace them, and that governments could implement tax incentives to encourage the hiring of human workers. “The future is so much subject to our choice and agency, it is all the more important that we have this discussion,” he concluded.