Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Google Deepmind co-founder and CEO Demis Hassabis, Mobile World Congress, the largest annual collection of Telecom industry, Barcelona, Spain, 26 February 2024.
Barrena Pau | AFP | Getty pictures
London – Artificial intelligence that can adapt to people in any assignment is still a way – it is just a matter of time before it becomes a reality. Google Deepmind.
Speaking at a briefing in London offices on Monday, Demis Hassabis said artificial general intelligence (AGI), which will begin to appear in the next five or 10 years, more intelligent or smarter than these people.
“I think today’s systems are very passive, but there are many things they could not.
As a Hassabis AGI, “a system that can show all the complex capabilities that people can do.”
“We are not there yet. These systems are very impressive in certain things. But there are other things that they are still unable to do and have many research work before,” Hassabis said.
Hassabis is not alone in suggesting that it will take a while for AGI. CEO of Chinese technological giant last year Baidu Robin Lee said he saw AGI “It’s more than 10 years“Some of some peers take some peers from this progress in the shorter.
Still go for a while
Hassabis’s forecast pushes the schedule for a later life of AGI’s industry compared to peers.
In January, CNBC CNBC told CNBC in the World Economic Forum in Davos, in the World Economic Forum in Davos, “Better all people in almost all positions in almost all positions” in January.
Other technological leaders see AGI coming soon. Ciscon’s chief crop worker Jeetu Patel thinks that this year is a chance we can see an example of the AGI. “There are three main stages,” Patel told CNBC in the interview of CNBC in Barcelona in Barcelona in Barcelona.
“Currently, there is a basic AI we have experienced. Then there is artificial general intelligence that cognitive opportunities meet people. Then there is something they call super management,” he said.
“I think you will see the meaningful evidence of AGI to be in the game in 2025. He added that there are years. “I think the super title is in the best case, it’s going out for a few years.”
Artificial super intelligence or ASI, after AGI, is expected to pass through the future and human intelligence. But when such a progress took place, “No one knows,” Hassabis said on Monday.
Last year, TESLA Director General Elon Musk said that AGI will be available until 2026, Openai CEO is Altman Such a system can be developed in “reasonable close.”
What’s up to reach AGI?
Hassabis is to achieve today’s AI systems of the main problem by obtain artificial intelligence to understand the real world.
It was possible to develop systems that can develop problems and complete the autonomy in the realm of the games – as Go to the Complex Strategy Board Game – This kind of technology is to the real world.
“Question, planning ideas and agency behavior, planning and reasoning, then working in the real world, we can summarize the world,” Hassabis said. “
“And I think we have made good progress with world models over the past few years,” he said. “So the question is now, what is the best way to combine with this planning algorithms?”
Director General of Google’s Cloud Computing Section Hassabis and Thomas Kurian, the so-called “multifarus” AI systems are a technological advancement that won a lot of traction behind the stage, he said.
Hassabis said he had been doing a lot to reach this stage. An example of example, Starcraft is the purchase of AI agents to understand how to play the popular strategy game.
“We have done a lot in the past, agents have done a lot with something in the league of society or agents, and they can cooperate because they can fight.”
“When I think about the agent communication agency, this is what we do to allow an agent to express himself … What are your skills? What tools do you use?” The Kurian said.
“These are all the elements you need to ask an agent, then other agents can contact him after this interface,” he said.
