Google DeepMind CEO Predicts Artificial General Intelligence Within a Decade

London, – Artificial General Intelligence (AGI)—AI capable of matching or surpassing human intelligence—could emerge within the next five to ten years, according to Demis Hassabis, CEO of Google DeepMind. Speaking at a press briefing in London, Hassabis acknowledged that today’s AI systems are impressive but lack critical capabilities needed to achieve full AGI.

AI’s Progress Toward AGI

Hassabis defines AGI as a system capable of exhibiting all the complex cognitive abilities of humans, including reasoning, planning, and understanding real-world context. While today’s AI models excel in specialized tasks, they are still far from achieving true general intelligence.

“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,” Hassabis said.

The prediction of AGI’s arrival within a decade contrasts with estimates from other tech leaders. For example:

• Dario Amodei, CEO of AI startup Anthropic, suggested that an AI system better than most humans at almost all tasks could emerge in the next two to three years.
• Jeetu Patel, Chief Product Officer at Cisco, believes that meaningful evidence of AGI could surface as early as 2025.
• Elon Musk and OpenAI CEO Sam Altman have also made similar forecasts, with Musk predicting AGI by 2026 and Altman stating that it could happen in the “reasonably close-ish future.”

Challenges in Achieving AGI

Despite rapid advancements in AI, reaching AGI remains a complex challenge. Hassabis highlighted three major hurdles:

1. Contextual Understanding: AI still struggles to interpret and reason about the real world, making generalization difficult.
2. Agentic Behavior: AI needs to develop decision-making capabilities similar to humans, which requires breakthroughs in planning and reasoning.
3. Multi-Agent Systems: DeepMind is focusing on AI agents that can cooperate, compete, and communicate, similar to how humans interact.

“We’ve done a lot of work on that with things like Starcraft in the past, where you have a society of agents. They could be competing, they could be cooperating,” Hassabis explained.

DeepMind has made progress in developing world models—AI systems that can understand their environment and make predictions. The next step is combining these models with advanced planning algorithms to bring AGI closer to reality.

Beyond AGI: The Race Toward Artificial Superintelligence

Once AGI is achieved, the next frontier is Artificial Superintelligence (ASI)—AI that surpasses all human cognitive abilities. However, Hassabis cautioned that “no one really knows” when ASI will emerge.

With Big Tech investing billions in AGI research, the race to develop human-level AI is intensifying. Whether AGI arrives in five years or a decade, one thing is clear—AI is evolving at an unprecedented pace, reshaping the future of technology and society.