Concerns about the risks posed by artificial intelligence continue to be a topic of debate in the tech world. Elon Musk believes the benefits of AI outweigh the risks. Many experts debate how risky “super-intelligent” AI can be.
Elon Musk: Artificial intelligence threatens humanity with a 20 percent chance
Speaking at a conference earlier this month, Elon Musk estimated that there is a 20 percent chance that advanced artificial intelligence poses an existential threat to humanity. But Roman Yampolskiy, an AI researcher, said Musk had underestimated the risks.
Yampolskiy argues that the “probability of apocalypse” from AI is much higher than Musk’s estimate. He puts the risk at almost 100 percent. According to Yampolskiy, this is because super-intelligent machines are getting to the point where they cannot be controlled by humans.
He said that because we cannot control AI, it is likely to harm humanity in ways we cannot foresee or prevent. Other experts shared similar concerns.
Geoff Hinton, a pioneer of neural networks, said that a super-intelligent AI raised by humans could learn to manipulate humanity in light of the examples we have given. Researchers have also found that once AI models have learned the concept of deception, they are difficult to reverse.
While Musk acknowledges the dangers, he believes that AI should still be developed because of its benefits. But Yampolskiy disagrees, arguing that uncontrolled superintelligence from any source carries serious risks.
{{user}} {{datetime}}
{{text}}