During the Nvidia's annual GTC conference on Tuesday, where the company discovered New Blackwell Ultra The family of the chips he (later this year), the high executive of he, he explained the relationships he thinks that human beings will have with him superintelligent.
Even if it progresses to the point that it is more intelligent than humans, capable of complex reasoning, understanding and problem solving, the chief of the meta scientists, Yann Lecun, says people will still be the boss of him.
Lecun sat for A fire conversation With the head of Nvidia Bill Dalli and discussed the possibility of replacing him. In response to Dally's statement that “he is not replacing people, essentially giving them energy means,” Lecun said that while he may have the potential to do so, people would never allow him to win the upper hand.
The relationship between people and the next one will be that of a boss and a staff of employees, 64-year-old Lecun predicted. Person person will have a team of that supervisor behind them.
Connected: He's Head of Meta is in the middle of a public clash with Elon Musk: 'Secret impedes progress'
“I mean, basically our relationship with the future systems of him, including supervision, is that we will be their boss,” Lecun said. “We will have a staff of supervisory, beautiful people, kind that work for us. I don't know about you, but I like to work with people who are smarter than me. Thing is the biggest thing in the world.”
Yann Lecun. Photography: Nathan Laine/Bloomberg through Getty Images
Lecun also said at the Nvidia conference that he does not believe in a “catastrophe scenario”, where a supervisor and powerful wants to take over society.
On the contrary, he believes that people will slowly and consistently develop oversight so that his sole purpose is to fulfill the tasks given by man. In short, even the smarter than people will have to respond to people.
Is a message he has said before. In one May 2024 posting On X, Lecun wrote that that superintelligent “will be a diligent solution of the problem that serves us.”
The display of supervision will not be an event.
We have nothing close to a project for superintelligent systems today.
At one point, we will come up with an architecture that can take us there.The design will begin with the level of intelligence of… https://t.co/vizphnsa0f
– Yann Lecun (@ilecun) May 31 2024
Openai Director General Sam Altman described a similar vision in his post on September blog titled “Age of intelligence“Altman wrote that one day, everyone will have a” personal team of he “consisting of experts in various fields that treat tasks like coordinating medical care.
However, supervision carries risks. Openai wrote in a December 2023 research newspapers This “an essential challenge” of supervision is that human beings will have to oversee it “much smarter than them”, leaving developers to ensure that even the advanced he remains safe and useful to humans. Openai's researchers predicted that supervision could take place within the next decade.
Lecun can be sure that people will prevent the advanced from taking on the jobs, but not all industry leaders are just as safe. Dario Amodei, for example, the general director of the 61.5 billion main dollars of the anthropic start, said last week that he could write any line of codes for software engineers within the next year.