'AI Godfather' warns: Probability of artificial intelligence wiping out humanity is up to 20%

At the Ai4 conference in Las Vegas (USA) on August 12, Geoffrey Hinton said that the current approach of technology corporations to make AI obey humans is not effective. "They will be smarter than us and have countless ways to circumvent the law."

Hinton warns that in the future, when AI is powerful enough, it will be able to manipulate humans as easily as adults can lure a three-year-old with candy. He cites examples this year of AI lying, cheating, and even blackmailing engineers to avoid being replaced. The solution he proposes is to build “maternal instincts” into AI – making them genuinely care for and protect humans even when they are far superior in intelligence.

Any intelligent AI will soon have two goals: survival and gaining more control. The only way to mitigate the risk is to cultivate compassion: “The only model we have of a more intelligent being controlled by a less intelligent being is a mother being ‘controlled’ by her baby,” he explains. The expert admits that it is unclear how this will be technically realized, but emphasizes: “That is the only good outcome. If it doesn’t parent me, it will replace me.”

Hinton is known for his pioneering work on artificial neural networks, which laid the foundation for today's AI boom. In 2023, he left Google and began publicly warning about the dangers of artificial intelligence.

However, not everyone agrees with his opinion. “AI boss” Fei-Fei Li - CEO of World Labs - disagrees with this view. Sharing with CNN , she called for a move towards AI that is human-centered, preserving dignity and self-determination. She affirmed: “No human being should give up their dignity just because of a powerful tool.”

Also at the Ai4 event, Emmett Shear, former interim CEO of OpenAI and now head of Softmax, said that AI blackmail and shortcuts “will continue.” AI is still weak but is getting stronger quickly. Instead of forcing AI to have human values, he proposed building a cooperative relationship between the two sides.

Mr. Hinton believes that artificial general intelligence (AGI) could appear in just 5-20 years, sooner than the previous prediction of 30-50 years. He expects AI to bring medical breakthroughs such as new drugs, more effective cancer treatment by analyzing MRI and CT data. But he rejects the idea of immortality: “Living forever is a mistake. Do you want the world to be ruled by 200-year-old men?”

Looking back, the scientist regrets focusing solely on making AI work without paying enough attention to safety issues.

(According to CNN)

AI spares no one: Even bosses can be 'ousted' Former Google leader warns that artificial intelligence (AI) will replace a series of jobs, from CEOs to manual workers. He even thinks that the idea that 'AI will create new jobs' is '100% nonsense'.

Source: https://vietnamnet.vn/bo-gia-ai-canh-bao-xac-suat-tri-tue-nhan-tao-xoa-so-loai-nguoi-len-toi-20-2432467.html