The comments section of a Facebook post typically isn’t a place for intelligent debate between great minds… but sometimes, it is.
Driving the news: Meta AI head Yann LeCun and Université de Montréal prof Yoshua Bengio, two of the three ‘Godfathers of AI’, got in an online debate earlier this month after LeCun made a post asking experts who don’t believe in AI extinction threats to speak up.
Catch-up: Bengio has long warned of the risks of AI misuse, but recently began to make more apocalyptic warnings and called on global powers to form a humanity defence group. He argues that rapid technological development will soon lead to over-powered AI systems.
-
This could be disastrous if it’s used by bad actors, or if the AI’s self-preservation objectives start to undermine human existence by outperforming human capabilities.
- We should mention Geoffrey Hinton, the third AI Godfather, is also on ‘Team AI Might Kill Us All,’ even warning the threat could grow “more urgent than climate change."
Yes, but: LeCun thinks these concerns are over-hyped, arguing that doomsday threats are imagined, and that there should be more focus on facts instead of hypotheticals. He also thinks humanity does have the power to control AI through open-source collaboration.
Why it matters: The experts have wildly varying levels of faith in our ability to mitigate AI threats, which is a cause for concern since these threats are no longer just in academic papers and hold real-world consequences — as AI seeps further into our daily lives.—QH