Elon Musk warns: there is a 20% chance that artificial intelligence will ‘end humanity’

The long-term impact of artificial intelligence In human life it is still a mystery. While this technology has come to solve many everyday problems, the debate is whether it may or may not exterminate civilization convene more and more experts. For Elon Musk, this hypothesis has 20% chance of happening.

This was stated by the CEO of Tesla, the social network X and SpaceX in statements to the site BusinessInsider that the risk that an artificial intelligence destroying humanity is between 10 and 20%.

However, it is important to note that Musk did not mention the risk of the said technology turning against humans and his prognosis is not that discouraging, as he stated that the possible benefits will make up for it any negative scenario.

During “The Great AI Debate,” Musk compared AI development to raising a child: “It’s almost like raising a child, but one that’s like a super geniuslike a child with divine intelligence, and it is important how the child is raised. “One of the things that I think is incredibly important for AI safety is to have AI that is extremely curious and truth-seeking.”

Despite his repeated criticism of this new technology, which he has called for regulation on several occasions, Elon Musk has never revealed how he performed the calculations. However, everything indicates that he would have used a variable called P(doom) which, unlike classical mathematics, does not require any specific method of investigation.

How to determine the real danger of artificial intelligence

The formula began as a joke on tech forums to describe the likelihood that artificial intelligence will destroy humanity. This informal variable without scientific basis seeks to explain If there is any chance that humanity will be annihilated for a specific invention.

Therefore, the variable P(doom) officially means “probability of doom” and, as the name suggests, refers to the chances of artificial intelligence causing an apocalyptic scenario.

The scale goes from zero to 100and the higher the score, the more convincing it will be Artificial intelligence is not just willing to eliminate humanityif necessary, but will actually be able to complete this task.

Because of its informality, it is difficult to establish a figure whose extinction is almost certain. Despite this, forecasts usually range from 5% to 50%in the first case it is quite probable that artificial intelligence will put an end to human existence as we know it, while the second represents an exaggeratedly high risk.

A large number of scientists and industry experts have declared that it is a real possibility; however, others have indicated the opposite.

One of the last is Michio Kakuknown as “the godfather of artificial intelligence”, who said that artificial intelligence is overrated, especially ChatGPT, which for him is nothing more than a “glorified recorder”.

So far, most figures in the field of artificial intelligence have volunteered some numbers to bring some peace of mind to humanity.

The co-founder of Ethereum, Vitalik Buterinwho warned about the existential threat of AI, is one of the least dangerous: it only has 10%.

OpenAI superalignment co-head Jan Leike puts it between 10 and 90, while Center for AI Safety director Dan Hendrycks recently updated his number from 20 to over 80. Even Elon Musk said it’s 20 or 30 .

Meanwhile, a recent survey of AI engineers found that their average p(doom) score was 40.