Pioneers of AI argue for open models despite safety concerns

At the Ai4 conference in Las Vegas, Nobel laureate Geoffrey Hinton, World Labs CEO Fei-Fei Li, and Coursera co-founder Andrew Ng spoke in favor of keeping AI open. Ng argued against gatekeepers, promoting openness and multiple providers to ensure competition and access. Hinton distinguished between open source software and open-weight models, expressing concerns about the latter being misused, but acknowledged that the battle is lost and open-weight models are here to stay. Ng worried about competitiveness, noting that cheaper Chinese open-weight models could influence the developing world and called for American open-source AI. Li cautioned against a dichotomy of complete openness versus closedness, advocating for nuance and citing examples like nuclear physics and the Human Genome Project, where different layers have different levels of openness. All three agreed that some regulation is necessary, with Hinton emphasizing the need to steer AI development towards helping people.