Skip to content
#AI

The ‘godfather of AI’ reveals the only way humanity can survive superintelligent AI (AI column)

#AI #Engineering #Science #Software

Jeffrey Hinton, a Nobel laureate, has been vocal about the risks of AI for the past two years. And here it is in August. spoke At the Ai4 conference in Las Vegas with a radical proposal for the security of artificial intelligence. He suggests embedding "mother instincts" into AI systems so that they take care of people, even if they become superintelligent. I thought his warnings were reasonable, so I decided to take a look, and when he changed the concept from being a techno-optimist, developing technology. As for the chronology of the last few years, it looked like this. Do. 2023 He spoke of AGI as a distant prospect. (30-50 years)

  • In March. 2023 The AGI may appear in less than 20 years
  • In May. 2023 Leaves Google for Open Discussion of AI Risks (later 10 company)
  • In 2024-2025 He reinforced warnings and proposed the concept of “mother instincts.”

In fact, his key warnings sound like this. There is a risk of extinction of mankind, where 10-20% probability that AI will destroy humanity Superintelligent AI may appear through 5-20 years (Forecasts were previously conservative. and) Current attempts by techno bros to keep AI subservient are doomed to failure because AI will be much smarter than humans.

But Jeffrey has the concept of using a mother-child model - the only known example of a less intelligent being controlling a more intelligent one. He believes that an AI with maternal instincts will not want to get rid of this quality, because it will not want people to die.

#AI #Engineering #Science #Software