Top OpenAI Scientist Transitions to Meta's AI Team

Jason Wei, previously a researcher with OpenAI, is set to join the superintelligence division at Meta, as revealed by insiders familiar with the development.

Wei has gained recognition for his contributions at OpenAI, focusing on o3 models and deep learning frameworks. Before his tenure with OpenAI, he was part of Google's team, where he developed methodologies for AI to tackle intricate problems sequentially. At OpenAI, Wei developed a strong interest in using reinforcement learning to train AI systems, a field that is quickly gaining traction and is of particular interest to Meta's newly formed superintelligence squad.

Confidential sources also disclosed to WIRED that Hyung Won Chung, another researcher from OpenAI, will be accompanying Wei to Meta. Both individuals' Slack accounts at OpenAI have reportedly been disabled. Requests for official comments from OpenAI, Meta, as well as Wei and Chung, were not immediately met with responses.

Chung reportedly engaged in many of the same research initiatives as Wei, including significant projects involving o1 models at OpenAI. His work emphasizes logical frameworks and agents, aligning with the skills sought after by Meta’s team. Records indicate that Chung and Wei concurrently worked at Google and joined OpenAI around the same time.

It appears that Meta has been strategically recruiting teams of researchers familiar with one another’s methodologies to strengthen its superintelligence initiative. This mirrors past strategies where Meta acquired several AI professionals who had collaborated on significant projects at their previous employers, such as the group moving from Google.

Meta's ongoing hiring spree targets elite AI talent, evidenced by recent internal movements. Last month, WIRED reported that Meta CEO Mark Zuckerberg circulated a confidential memo detailing a revamped AI strategy, spotlighting several new recruits, many plucked from OpenAI's ranks.

The intense competition for AI expertise shows no signs of easing. In response, OpenAI has bolstered its own team, recently acquiring four senior engineers from other tech giants, including Tesla and Meta.

In a reflective social media post, Wei elaborated on the personal insights he gained from his work in reinforcement learning, emphasizing the value of moving beyond mere imitation to achieve novel outcomes by embracing unique strategies and navigating the dualities of risk and reward.