Musk says there's a 20% chance of artificial intelligence destroying humanity, but it's still worth the risk

News from this site on April 1, Elon Musk said at the "Artificial Intelligence Debate" seminar held at the Abundance Summit earlier that even if artificial intelligence technology has 1/5 The possibility is that it could pose a threat to humanity, but the benefits outweigh the risks and it's still worth the risk to pursue research and development.

Musk re-evaluated his previous risk assessment of artificial intelligence, saying at the seminar: "I think artificial intelligence has the potential to end human civilization. I probably agree. Geoffrey Hinton's point of view, The probability is about 10% to 20%." But he added: "I think the positive possibility scenario is greater than the negative possibility scenario." Musk did not mention to how he calculated that risk.

This site noticed that in November last year, Musk said that the possibility of this technology getting out of control was "not zero", but at that time he did not assert that out-of-control artificial intelligence would destroy mankind.

While Musk has been an advocate for regulating artificial intelligence, last year he launched a company called xAI to promote the power of artificial intelligence. xAI is a competitor to OpenAI, the company Musk co-founded with Sam Altman, but Musk resigned from the OpenAI board in 2018.

At the summit, Musk estimated that by 2030, digital intelligence will exceed all human intelligence combined. While he still insists the potential positives outweigh the negatives, Musk acknowledged the risks that AI will pose to the world if it continues on its current trajectory in some of the most direct terms he has ever used in public.

Speaking about artificial intelligence at a Silicon Valley event on March 19, Musk said, "I think a very important thing for the safety of artificial intelligence is to make it as realistic and curious as possible." He raises an artificial intelligence, which is almost like raising a child, but it is a super genius, a godlike child - the way you raise it is crucial. We must remain highly vigilant about the safety of artificial intelligence, make it as real as possible, and make it full of curiosity. This is a very critical point. Allowing it to pursue truth and remain curious will ensure that it does not go off track and remains curious and pursuing truth.

Musk said his "ultimate conclusion" on the best way to achieve safety in artificial intelligence is to let it grow in a way that forces its honest implementation. This approach can not only ensure the safety of artificial intelligence, but also promote its further development.

"Don't force it to lie, even if the truth is unpleasant," Musk said of ways to make artificial intelligence safe. "That's very important. Don't force it to lie."

The Independent Report reported that researchers have found that if some artificial intelligence learns to attack humans, current artificial intelligence security measures cannot be used to defend against such deception. More disturbingly, the researchers add, it could be possible for AI to learn to deceive on its own, rather than being influenced by being specifically taught lies. Therefore, the researchers call on AI learners to be vigilant on their own, rather than being tricked by deliberate teaching.

The above is the detailed content of Musk says there's a 20% chance of artificial intelligence destroying humanity, but it's still worth the risk. For more information, please follow other related articles on the PHP Chinese website!