AI Will Be as Good as Humans at Any Task in Ten Years, Google DeepMind CEO Says

Google DeepMind CEO Demis Hassabis has stated that strong AI (Artificial General Intelligence, AGI), which is as good as or better than humans, will be developed in the next five to ten years. He is absolutely confident in the feasibility of creating AGI and believes that it is only a matter of time before this problem is solved. Hassabis defines AGI as “a system that can demonstrate all the complex capabilities of a human being.”

Image source: Pixabay

«”I think today’s systems are very passive, there are still a lot of things they can’t do. But I think over the next five to 10 years, a lot of these capabilities will start to come to the fore, and we’ll start to move toward what we call artificial general intelligence,” says Hassabis. He’s not alone in his view – last year, Robin Li, CEO of Chinese tech giant Baidu, said AGI was “more than 10 years away.”

Others at Hassabis are more optimistic. Anthropic CEO Dario Amodei is confident that an AI model that is “better than almost all humans, at almost all tasks” will emerge in the “next two to three years.” Cisco Chief Product Officer Jeetu Patel believes AGI could be built as early as this year. Tesla CEO Elon Musk has predicted that AGI will likely be a reality by 2026, and OpenAI CEO Sam Altman believes such a system could be developed in the “reasonably near future.”

Hassabis went further in his predictions, suggesting that AGI will be followed by Artificial Super Intelligence (ASI), which will surpass humans in all areas of activity. However, “nobody really knows” when such a breakthrough will occur, he admitted.

According to Hassabis, the main challenge in creating AGI is bringing today’s AI systems to the level of understanding the context of the real world. “The question is how quickly can we generalize the ideas of planning and agent behavior, planning and reasoning, and then apply them to the real world, complementing them with things like world models that can understand the reality around us,” he explained.

According to Hassabis, so-called multi-agent AI systems have recently attracted more and more attention from developers. As an example, he cited DeepMind’s research into training AI agents to play Starcraft: “We’ve done a lot of work on this, for example in the Starcraft project, where we had a community of agents or a league of agents that could both compete and cooperate.”

Senior managers and leading AI developers agree on one thing: they do not see the near future of humanity without omniscient, omnipotent, and omniscient AI systems that they are confident will surpass humans in every field of activity. Apparently, they think that they and their loved ones will not be affected by such an invasion of AI locusts “blowing out the sky.”