LONDON: Artificial General Intelligence (AGI) — an advanced form of artificial intelligence capable of performing virtually any intellectual task at a human level — could become a reality within the next three to four years, according to Google DeepMind CEO Demis Hassabis.
In a recent blog post, Hassabis described the current period as a pivotal moment for humanity and stressed the need for urgent global action to ensure that increasingly powerful AI systems are developed and used safely for the benefit of society.
He defined Artificial General Intelligence (AGI) as an AI system capable of matching or even surpassing human abilities across a broad range of tasks, unlike today’s AI models, which excel mainly at specific tasks.
Hassabis urged governments around the world to establish an international oversight body to monitor the development of advanced AI technologies. He warned that the current focus on competition among AI companies and nations risks sidelining critical safety concerns.
According to Hassabis, the rapid pace of AI development makes robust governance and safeguards essential, particularly as future systems are expected to become more autonomous and capable of improving themselves with minimal human intervention.
Despite his optimism about AGI, Hassabis acknowledged that predicting the exact capabilities of future AI systems remains extremely difficult, even for leading researchers in the field.
Speaking in March 2025, he said AGI would be able to perform the full range of complex tasks that humans can do. However, he noted that current AI systems, while highly impressive in specialized areas, still struggle with many aspects of real-world reasoning and general understanding.
Looking further ahead, Hassabis said the development of Artificial Superintelligence (ASI)—AI that significantly surpasses human intelligence—is expected to follow AGI. However, he emphasized that no one can accurately predict when such a breakthrough might occur.
He identified one of the biggest challenges in achieving AGI as enabling AI systems to understand the real-world context in which they operate. While substantial progress has been made in recent years, he said researchers must still solve how to effectively combine advanced planning algorithms with broader reasoning capabilities to create truly general-purpose intelligence.
About The Author
