AI pioneers Hinton, Li and Ng debate the future of open models

At the Ai4 conference in Las Vegas, three influential AI researchers — Nobel laureate Geoffrey Hinton, World Labs CEO Fei-Fei Li, and Coursera co-founder Andrew Ng — addressed how open AI development should be, amid growing industry concerns about safety.

All three expressed worry that a small number of major companies could end up controlling the pace of AI progress, drawing comparisons to how Apple and Google dominate mobile operating systems. Ng said he did not want "gatekeepers" limiting access to AI and argued for maintaining multiple competing providers and models.

Differing views

Hinton distinguished between open-source software, where code is available for inspection, and open-weight models, where a trained model's parameters are released publicly. He said he had previously opposed open weights because they make it easier to misuse expensive foundation models for harmful purposes such as cyberattacks, but acknowledged that battle had already been lost, since open-weight models are now firmly established.

Ng focused on competitive dynamics, arguing that whoever builds the cheapest model gains an advantage. He warned that if China's open-weight models achieve wide adoption across Asia, Africa, and the developing world, they could shape how billions of people encounter ideas about democracy and human rights, especially if U.S. open-source AI struggles to keep pace due to lobbying and fear-mongering.

Li rejected framing the issue as a binary choice between full openness and full closure, comparing it to nuclear physics, where research papers are published openly while uranium is tightly regulated. She said different layers of the AI ecosystem could operate at different levels of openness, citing the Human Genome Project as an example of successful public-private collaboration.

Despite their differences, all three agreed that some degree of regulation would be necessary to keep AI development aligned with the public interest.