Hinton, Ng, Li defend open AI models at Ai4 conference

At the Ai4 conference in Las Vegas, three of AI's most influential figures — Nobel Prize winner Geoffrey Hinton, World Labs CEO Fei-Fei Li, and Coursera co-founder Andrew Ng — made a unified stand for openness in AI, even as safety concerns push major labs toward controlled releases. Hinton distinguished open source software (where code is inspectable) from open-weight models (where trained parameters are released), admitting he was "against open weights" because they make it easy to fine-tune powerful foundation models for malicious uses like cyber attacks. However, he acknowledged the battle is lost: "It's too late." Ng framed the issue around gatekeepers, warning that a handful of dominant companies could restrict access and innovation. He emphasized competition and warned that China's open-weight models could gain traction across Asia and Africa, making American open-source AI a matter of soft power and cost efficiency. Li pushed back against framing the debate as a false dichotomy between complete openness and complete closure, urging a more nuanced path.

Despite their tactical disagreements, all three agreed that AI progress should not be controlled by a few major labs. Hinton said AI would boost productivity, education, and healthcare, and defended legitimate worry about existential risks from being labeled fear-mongering. Ng called for promoting openness, saying, "I want it to be in everyone's hands." The discussion reflects the industry's growing tension between safety-driven closed development and the democratic benefits of open-weight distribution — a tension that will shape AI governance, competitive dynamics, and global influence for years to come.