How AI Is Being Trained to Do Your Job

The rapid growth of AI-training gig work, exemplified by the startup Mercor, highlights a new labor category where human experts teach large language models to perform specialized white-collar tasks. Companies like Mercor act as intermediaries, hiring professionals—ranging from linguists to creative writers—to refine AI outputs, correct linguistic nuances, and eliminate repetitive "AI-isms." While this work offers immediate income, it presents significant ethical and professional dilemmas, including concerns over intellectual property rights and the irony of workers accelerating the automation of their own careers. Former contractor Carolina Perez-Sanz, a speech pathologist, observed the AI’s rapid improvement firsthand before becoming disillusioned by the diminishing pay and the perceived societal impact of the technology. Despite legal challenges and data privacy concerns, the demand for human-in-the-loop training continues to surge, underscoring the relentless drive to make AI models capable of replacing human expertise.