At the India AI Impact Summit 2026, DeepMind CEO Demis Hassabis stated that current AI systems remain far from achieving Artificial General Intelligence (AGI). He highlighted limitations such as lack of continual learning, weak long-term planning, and inconsistent reasoning, predicting that true AGI is at least five years away.
Demis Hassabis, CEO and co-founder of Google DeepMind, addressed the India AI Impact Summit 2026 in New Delhi, emphasizing that today’s AI systems, while powerful, are not yet capable of achieving general intelligence. He explained that current models excel in coding, mathematics, and games where outputs are verifiable, but fail in subjective domains requiring nuanced reasoning.
Hassabis described today’s AI as “jagged intelligence”—strong in complex tasks yet weak in basic reasoning. He stressed that continual learning, coherent long-term planning, and consistent reasoning are missing pieces that must be solved before AGI can emerge. According to Hassabis, the timeline for achieving AGI is at least five years away, underscoring the need for sustained research and innovation.
Key Highlights
-
Current AI Strengths: Coding, mathematics, and games with verifiable outcomes.
-
Limitations: Lack of continual learning, weak long-term planning, inconsistent reasoning.
-
Concept of “Jagged Intelligence”: AI excels in complex tasks but struggles with basic reasoning.
-
AGI Timeline: Hassabis predicts at least five more years before true AGI.
-
Broader Vision: AI is entering a “golden era” of scientific discovery, but general intelligence remains elusive.
Sources: The Economic Times, India Today, YourStory, Digit