Claims by Meta's chief AI scientist, Yann LeCun, that AI won't wipe out humanity are dangerous and wrong according to one of his fellow AI godfathers.
"Yann and others are making claims that this is not a problem, but he doesn't know — nobody knows," Yoshua Bengio, the Canadian computer scientist and deep learning pioneer, told Insider Thursday. "I think it's dangerous to make these claims without any strong evidence that it can't happen."
LeCun, the storied French computer scientist who now leads Meta's AI lab, sparked a furious debate on X last weekend after accusing some of the most prominent founders in AI of "fear-mongering" with the ultimate goal of controlling the development of artificial intelligence.
LeCun argued that by overstating the apparently farfetched idea that AI will wipe out humans, these CEOs could influence governments to bring in punitive regulation that would hurt their competition.
"If your fear-mongering campaigns succeed, they will inevitably result in what you and I would identify as a catastrophe: a small number of companies will control AI," LeCun wrote.
Bengio, who once worked with LeCun at Bell Labs and was co-awarded the Turing Award with him in 2018 for their work in deep learning, told Insider that LeCun was too dismissive of the risks.
"Yann himself agreed that it was plausible we would reach human-level capabilities in the future, which could take a few years to a few decades and I completely agree with the timeline," he said. "I think there's too much uncertainty, and the stakes are too high."
Bengio has said in the past that current AI systems are not anywhere close to posing an existential risk but warned things could get "catastrophic" in the future.
AI's leading lights are unlikely to come to a consensus any time soon.
Andrew Ng, cofounder of Google Brain, said this week that big tech was over-inflating the existential risks of AI to squash competition from the open-source community.
As AI's biggest names subsequently began piling in, Yann went on to call out his fellow Turing Award winners Geoffrey Hinton and Bengio in another post.
In a response to Hinton, who has claimed AI poses an extinction risk, LeCun wrote on X: "You and Yoshua are inadvertently helping those who want to put AI research and development under lock and key and protect their business by banning open research, open-source code, and open-access models."
Bengio did warn governments need to ensure they weren't only listening to tech companies when formulating regulation.
"In terms of regulation, they should listen to independent voices and make sure that the safety of the public and the ethical considerations are at center stage," he said.
"Existential risk is one problem but the concentration of power, in my opinion, is the number two problem," he said.
