John Schulman

• 1. Introduction

  ›

  John Schulman was in seventh grade when he became fixated on BattleBots. The television program featured combat between remote-controlled robots, and it set off what he described as his first experience of self-directed learning. He read extensively across technical subjects hoping to design a superior robot. The robot was never built.

  The habit of chasing hard problems through intensive reading never left him. It would eventually carry him from a physics competition circuit to the frontier of AI research. The specifics of how that habit shaped a career are worth following from the beginning.

• 2. Physics and Self-Study

  ›

  Isaac Asimov's science fiction was among the early reading that ran alongside Schulman's formal study of science and mathematics. By 2005, as a student at Great Neck South High School, he had earned a place on the US Physics Olympiad Team.

  In 2010, he graduated from Caltech with a degree in physics. Graduate work followed at the University of California, Berkeley, where he pursued a doctorate in electrical engineering and computer sciences under Pieter Abbeel. He was still completing that work when an opportunity arrived that would change the scale of everything he was pursuing.

• 3. Eleven Co-Founders

  ›

  Eleven researchers founded OpenAI in December 2015, with Sam Altman and Elon Musk as the organization's co-chairs. Schulman was among them, still finishing his doctorate. The other co-founders were Ilya Sutskever, Greg Brockman, Trevor Blackwell, Vicki Cheung, Andrej Karpathy, Durk Kingma, Pamela Vagata, and Wojciech Zaremba.

  Within the new organization, Schulman took charge of the reinforcement learning team. Reinforcement learning trains AI systems by rewarding desired behaviors rather than by giving them labeled examples of correct outputs. The team worked toward what would become ChatGPT, the conversational AI system that put OpenAI's name into wide public use. He has been referred to as the "architect" of ChatGPT, a designation that recognized the centrality of his team's methods to the project. His focus within that work shifted over the years, and the question he most wanted to pursue eventually required a different institutional home.

• 4. Toward Alignment

  ›

  In August 2024, Schulman announced he was leaving OpenAI for Anthropic. He stated that the move would allow him to deepen his focus on AI alignment and to return to more hands-on technical work.

  AI alignment addresses the problem of ensuring that AI systems pursue the goals their developers intend. The concern is that a sufficiently capable system might otherwise behave in unpredictable or harmful ways. Naming alignment as the primary motivation for leaving an organization he had helped found was a clear statement of where his priorities had settled.

  His time at Anthropic was brief. In February 2025, he announced he was joining Thinking Machines Lab as chief scientist. Thinking Machines Lab became his third institutional home in under a year. The Mark Bingham Award from UC Berkeley recognized him that same year for excellence in achievement by young alumni. It marked a first decade of work that shows no sign of being the last.