ChatGPT co-creator Shengjia Zhao to spearhead Meta’s Superintelligence Lab

The Superintelligence Lab, co-founded by Zhao, centralizes Meta’s cutting-edge AI research—especially its Llama models—and functions independently from FAIR, the company’s original AI division led by noted deep learning expert Yann LeCun

Meta Platforms has appointed Shengjia Zhao, a key contributor to the development of ChatGPT and GPT-4, as the chief scientist of its newly established Superintelligence Lab. The announcement was made by CEO Mark Zuckerberg via a post on Threads, highlighting the company’s strategic move to accelerate its ambitions in advanced artificial intelligence.

Zhao, previously a research scientist at OpenAI, played a leading role in building models like ChatGPT, GPT-4, and other smaller-scale AI systems such as GPT-4.1 and o3. He is now tasked with shaping the scientific vision and research roadmap of Meta’s Superintelligence Lab, reporting directly to Zuckerberg and Meta’s Chief AI Officer, Alexandr Wang. Wang was recruited from startup Scale AI following Meta’s significant investment in the company.

Meta intensifies AGI talent race

Zhao’s hiring is part of a broader trend of AI talent shifting from OpenAI to Meta in recent weeks, as the tech giant intensifies efforts to close the gap with competitors in the race for artificial general intelligence (AGI). Meta has been offering highly competitive compensation packages and forming alliances with startups to attract leading researchers in the field.

The Superintelligence Lab, which Zhao co-founded, is designed to centralize Meta’s most advanced AI research, including ongoing work on its Llama models. It operates independently of Meta’s long-established AI research division, FAIR, headed by renowned AI expert Yann LeCun.

Zuckerberg has publicly stated that Meta’s goal is to build “full general intelligence” and to release much of its work as open source. While the move has drawn attention and praise for promoting transparency, it has also sparked debate over the potential risks of widely accessible, powerful AI tools.