Google DeepMind CEO Demis Hassabis predicts that artificial general intelligence (AGI) capable of matching or exceeding human capabilities could emerge within the next five to ten years, signaling a profound shift in the AI landscape.
“I think over the next five to 10 years, a lot of those capabilities will start coming to the fore and we’ll start moving towards what we call artificial general intelligence,” Hassabis said during a briefing at DeepMind’s London headquarters.
The DeepMind chief’s forecast contrasts with more aggressive timelines from other industry leaders. OpenAI CEO Sam Altman and Tesla CEO Elon Musk, two contrasting personalities, share the same prediction of AGI arriving in just 2-3 years. Anthropic CEO Dario Amodei similarly believes AGI could emerge “in the next two or three years.” Meanwhile, Mustafa Suleyman, CEO of Microsoft AI, aligns more closely with Hassabis, suggesting AGI won’t be ready for at least 5-10 years.
Demis Hassabis says AI is “not quite there yet”
While current AI systems have demonstrated impressive performance in specific domains, Hassabis acknowledged significant limitations remain. “We’re not quite there yet. These systems are very impressive at certain things. But there are other things they can’t do yet, and we’ve still got quite a lot of research work to go before that,” he explained. He defined AGI as “a system that’s able to exhibit all the complicated capabilities that humans can.”
Hassabis identified real-world context understanding as the primary challenge to achieving AGI. While AI has mastered complex games like Go, transferring such capabilities to real-world applications presents greater difficulties.
“The question is, how fast can we generalize the planning ideas and agentic kind of behaviors, planning and reasoning, and then generalize that over to working in the real world,” Hassabis said.
Multi-agent AI systems represent a promising advancement, according to Hassabis and Google Cloud CEO Thomas Kurian. DeepMind has explored this concept through experiments with the strategy game “Starcraft,” developing “a society of agents, or a league of agents” that can compete or cooperate with one another.
