Mercor’s 60 Minutes moment puts the humans behind AI in the spotlight

Most people picture AI training as servers humming in a dark warehouse. Mercor would like you to picture a doctor at her kitchen table, grading a chatbot’s answer about drug interactions. The San Francisco company got the 60 Minutes treatment in an October 4, 2026, segment. The episode focused on its network of more than 100,000 freelancers who use their professional expertise to make AI models smarter. How the Mercor machine works Mercor sits between the biggest AI labs and a large pool of skilled contractors. Its clients include OpenAI and Anthropic. The freelancers come from fields ranging from medicine to finance. Their job is to give feedback that helps AI systems learn what a good answer actually looks like. The technical term is reinforcement learning from human feedback, usually shortened to RLHF. Think of it as a very demanding tutor marking up a student’s essays, except the student is a language model and the essays never stop coming. Mercor also supplies experts for model evaluations. These are structured tests that check whether a model’s output holds up to professional scrutiny before it reaches the public. Not everyone in the network works every week. Roughly 30,000 c...