According to multiple sources familiar with the matter, OpenAI researcher Jason Wei is joining Meta’s new super-smart lab.
According to his personal website, WEI studied Openai’s O3 and in-depth research models. After some time at Google, he joined OpenAI in 2023, where he engaged in thoughtful research involving training AI models to process complex queries step by step. In Openai, WEI becomes a “stubborn” who calls itself reinforcement learning, a way to train or perfect AI models with positive or negative feedback. It has become a promising area of AI research – a researcher hiring a dedicated researcher for its ultra-intelligent team.
Another Openai researcher, Hyung Won Chung, will also join Meta, a source told Wired. Multiple sources confirm that both WEI and Chung’s internal OpenAi Slack curves are currently deactivated. Openai, Meta, Wei and Chung did not immediately respond to Wired’s request for comment.
According to Chung’s personal website, Chung has collaborated with WEI in some projects at Openai, including in-depth research and Openai’s O1 model. The website says his research focuses on reasoning and agency. Chung also overlaps with WEI on Google and joins Openai at the same time based on WEI’s LinkedIn profile.
Several sources told Wired that Wei and Zhong had a close working relationship. Meta has previously worked on his new super smart lab, including a trio from Openai’s Swiss office and joined Google’s Chatgpt Maker.
Meta has been carrying out a poaching craze over the past month, providing up to $300 million in AI talent over four years. Wired reported late last month that FW CEO Mark Zuckerberg sent an internal memo to staff to develop new plans for the company’s AI efforts. It includes a super smart team of new employees, most of whom are recruited from Openai.
The recruitment madness shows no signs of slowing down, Openai has been fighting back. Just last week, Wired reported that Openai had recruited four senior engineers from Tesla, XAI and Meta.
Wei shared an article on social media on Tuesday reflecting on what he called “important lessons” and reinforcement learning taught him “how to live my own life.”
Wei writes that in life (and when building AI models), imitation is good, and you have to do it at the beginning. But “beating teachers requires going your own path, taking risks and rewarding from the environment.”
