New Delhi: Billionaire Elon Musk shared a three-year-old video in which he warned there is a “small chance” artificial intelligence might kill all of us.
In an old interview shared by the SpaceX founder and CEO, he said, “I think what’s important in training AI and growing an AI is to make sure that it is as truthful as possible and maximally curious”.
While sharing the video on Sunday, Musk wrote, “Yes”.
In a separate post, he said, “I’ve been sounding the alarm on AI for a long time.”
Growing concerns over AI
Musk’s repost of an old video comes amid growing concern about AI taking over humanity, re-emphasising his decade-long warnings regarding the existential risks of Artificial General Intelligence (AGI).
Last week, Jacob Coxon, who has worked as a safety researcher for OpenAI and Anthropic, posted on X that both platforms are not acting responsibly.
Coxon further said, “The people building AI earnestly believe that it could kill us all by the end of the decade”.
Musk also reiterated Coxon in the reposted video.
“Artificial intelligence is being smarter than the smartest human, at which point, you know, perhaps any invention is possible”, he said. “Let’s hope not. Because I think if that’s true, then I think it’ll probably foster humanity. But it is very important that one is trained to be honest, even if that truth is unpopular.”
Musk suggested three ways to make AI better. He urged platforms to make AI “truthful as possible”, “maximally curious” and “train it to be honest even when the truth is unpopular.”
Also read: ‘Chased, filmed, forced into races’—women bikers say Gurgaon case is common on Indian roads
‘We need to pace the frontier’
Anthropic CEO Dario Amodei also raised concerns over AI’s speed to adapt. He highlighted that this might help in the healthcare sector, but if these capabilities are not kept in check, it can pose a serious risk.
OpenAI CEO agreed with the statement.
“I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we’ve had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We’ll have more to share soon,” he posted on X.
X users reiterated Musk’s words.
One user wrote, “Truth seeking is probably the best argument thus far in controlling AI. However, even with truth seeking, there is an acknowledgment that AI could extinguish humankind”.
(Edited by Saptak Datta)