A star OpenAI engineer says he's starting a new team to chase artificial superintelligence

In a bold move, a key engineer behind OpenAI's innovative text-to-video technology, Sora, has announced the formation of a new team aimed at achieving artificial superintelligence (ASI). Will DePue, who played a significant role in developing Sora, revealed on social media that he is rejoining OpenAI after a break to collaborate with fellow engineers Troy Luhman and Eric Luhman. DePue described the initiative as an 'incredibly high-risk bet' with the potential to lead to groundbreaking advancements in AI. He emphasized the team's commitment to maintaining a small, focused group while remaining open to high-caliber researchers and engineers eager to explore uncharted territories in AI development. As of now, details regarding the newly formed team have not been disclosed by OpenAI, and the company has yet to respond to inquiries from media outlets. DePue, who began his academic journey in computer science at the University of Michigan, has publicly shared that he has taken a leave of absence from his studies. He initially joined OpenAI in July 2023 and has since reestablished his role at the organization. His contributions include training ChatGPT models and advancing the Sora project, which was officially launched in December. Notably, DePue's personal website highlights various innovative projects he has undertaken, including a browser-based implementation of GPT and an embedding dataset project named the 'Alexandria Index.' While there is limited information available about his colleagues Troy and Eric Luhman, both have made contributions to machine learning research. The trio previously collaborated on the Sora project, which debuted in February and saw its second iteration released in September. OpenAI has articulated its ambition of reaching artificial general intelligence (AGI) — a state where AI can emulate human reasoning. However, CEO Sam Altman has indicated that the organization is already setting its sights on superintelligence, a concept that could enable AI to reason and innovate at levels far beyond human capabilities. In a blog post earlier this year, Altman articulated the potential of superintelligent tools to revolutionize scientific discovery and significantly enhance global prosperity.