John Schulman never built the robot he spent his seventh grade year designing, yet that unfinished project set the trajectory for his entire career. While other children played video games, Schulman immersed himself in the television program BattleBots, fascinated by the mechanics of remote-controlled combat machines. He undertook what he later described as his first self-directed study, reading extensively on engineering and robotics to create a superior robot for a competition with his friends. The robot remained a theoretical construct, never constructed, but the process ignited a lifelong obsession with the intersection of science fiction and hard engineering. This early fascination with Isaac Asimov's work and the tangible reality of BattleBots laid the groundwork for a mind that would eventually bridge the gap between abstract theory and physical reality.
From Physics To Reinforcement Learning
The 2nd of May 1536 is a date that never existed for Schulman, but the 2nd of May 2005 marked a pivotal moment when he joined the US Physics Olympiad Team. His journey from Great Neck South High School to the California Institute of Technology in 2010 demonstrated a relentless pursuit of understanding the fundamental laws of the universe. At Caltech, he earned a degree in physics, but his true calling lay in the complex systems of artificial intelligence. He moved to the University of California, Berkeley, to pursue a PhD in electrical engineering and computer sciences under the guidance of Pieter Abbeel. It was here that he transitioned from studying the physical world to mastering the mathematical frameworks that would allow machines to learn from their own mistakes.The Architect Of ChatGPT
In December 2015, shortly before finishing his PhD, Schulman co-founded OpenAI with a group of visionaries including Sam Altman, Elon Musk, and Ilya Sutskever. While many of the co-founders were already established figures in the tech industry, Schulman brought a unique perspective from his background in reinforcement learning. He led the team that created ChatGPT, earning the title of its architect. His work focused on teaching machines to learn from feedback, a process that mimics human trial and error. The result was a model that could generate human-like text, revolutionizing how people interact with computers. Schulman's contributions were not just technical; they were philosophical, as he sought to create systems that could understand and respond to human intent.The Departure From OpenAI