ChatGPT-maker OpenAI's VP Researcher Jerry Tworek resigns; says: I am leaving to try and explore types of research that are hard to do at OpenAI, this is a...

Jerry Tworek, a veteran researcher and vice president at OpenAI, has announced his departure from the company after nearly seven years. Tworek led the development of OpenAI's groundbreaking o1 reasoning model and was instrumental in creating the technology behind GitHub Copilot and ChatGPT's coding capabilities.In a post on X, Tworek shared the memo he also sent to his team at OpenAI. He said his decision to leave was driven by a desire to pursue research avenues difficult to explore at OpenAI. "While I am leaving to try and explore types of research that are hard to do at OpenAI, this is a special company and a special place in the world that already has a forever place in history of humankind," he wrote.

Key architect behind OpenAI's reasoning model

Tworek joined OpenAI in 2019, when it was still a nonprofit research lab, and helped shape many of the company's most notable breakthroughs.

He led the team that developed the o1 model-a milestone that introduced a new approach to AI by training models to "think" through problems with reinforcement learning rather than relying on pre-training data.In August, the o1 model topped the leaderboard of LiveBench, showcasing state-of-the-art performance on challenging reasoning tasks. Tworek also led research on OpenAI Codex, the engine behind GitHub Copilot, and contributed to the programming capabilities of GPT-4 and the plugin integration in ChatGPT.

Researcher reflects on nearly seven years of AI breakthroughs at OpenAI

In his departure note, Tworek reflected on his tenure with fondness. He highlighted his work on "scaling RL on robots," training early coding models, discovering chinchilla scaling principles, and building teams that established the reasoning model paradigm. "I've developed many friendships, spent some nights at the office, contributed to and witnessed a healthy number of technological breakthroughs," he shared.Tworek expressed gratitude for the trust OpenAI placed in him over the years. "Those types of moments never feel right, but in a good and optimistic framing they can be a catalyst for something great," he wrote, signing off with: "Take great care and stay warm my dearest strawberries."