Doctors Train AI to Diagnose and Prescribe

Former Stanford emergency medicine instructor Dr. Alice Chiao is now training artificial intelligence chatbots to think and diagnose like medical professionals. She is part of a rapidly expanding economy where experts in various fields teach AI through reinforcement learning, a trial-and-error process of grading responses. This service industry is estimated to be worth at least $17 billion, with companies like Mercor managing contracts for thousands of experts.

Experts in medicine, law, finance, and even comedy can earn up to hundreds of dollars per hour. Dr. Chiao views her work not as replacing doctors, but as ensuring AI is safe and accurate, allowing physicians to focus more on patient care. Mercor, a San Francisco-based AI company, was recently valued at $10 billion, highlighting investor confidence in human feedback as a crucial part of AI development.

Mercor, co-founded three years ago, shifted from HR to AI training, leveraging its extensive resume database. The company now pays out over $1 million daily to thousands of experts. Its rapid growth and valuation suggest that services like AI model testing are becoming essential. Competitors like Scale AI, with significant investment from Meta, also operate in this space.

While AI can master technical tasks, more subjective areas like comedy prove challenging due to varied human humor. Mercor's CEO, Brendan Foody, believes AI's increased productivity will help society tackle major issues like cancer and climate change. Mercor's success positions its young founders among notable tech entrepreneurs, with Foody expressing that the rapid growth feels surreal.