Demis Hassabis predicts AGI in 5-8 years, sees new golden era transforming medicine and science

Demis Hassabis Sets Timeline for Artificial General Intelligence

Demis Hassabis, the Nobel Prize-winning CEO of Google DeepMind, has placed artificial general intelligence (AGI) on a definitive timeline, predicting its arrival within five to eight years2

2

3

3

. Speaking at the India AI Impact Summit 2026 in New Delhi, Hassabis declared that humanity is "starting to see what these systems can do," marking a watershed moment in AI development. His forecast positions AGI impact at ten times the magnitude of the Industrial Revolution, but unfolding at ten times the speed—compressed into a decade rather than a century4

4

. This timeline represents a recalibration from his earlier predictions, with the year 2026 now serving as a pivotal marker on the path to human-level AI systems.

Fixing AI's Jagged Intelligence Problem

Before AGI can materialize, Hassabis identifies a critical obstacle: what he terms "jagged intelligence." AI systems today exhibit dramatic inconsistencies, excelling at complex tasks like securing gold medals at the International Mathematical Olympiad while simultaneously failing elementary math questions2

2

. These systems are "very good at certain things, but very poor at certain things," a dichotomy that must be resolved before achieving true general intelligence. Hassabis pinpoints continual learning as one of the biggest gaps, noting that current models are "trained, fine-tuned, and then essentially frozen before being released into the world"3

3

. AI systems need the ability to learn from real-world experience, adapt to different situations, and personalize themselves to specific tasks—capabilities that remain elusive. Long-term planning presents another limitation, with models struggling to construct coherent plans spanning months or years as humans naturally do.

New Golden Era of Discovery Through AI

Beyond the AGI milestone, Hassabis envisions a transformative new golden era of discovery emerging within 10 to 15 years, characterized by what he calls "radical abundance"1

1

5

5

. Speaking on the Fortune 500: Titans and Disruptors of Industry podcast, he predicted that "medicine won't look like it does today," with AI enabling personalized treatments and curing major diseases. His vision extends to energy, where AI could unlock new materials to solve the energy crisis through fusion or solar breakthroughs, and ultimately to space exploration, allowing humanity to "travel the stars and explore the galaxy." This renaissance hinges on AI successfully bottling the scientific method to tackle the planet's most intractable problems. Hassabis points to AlphaFold, DeepMind's breakthrough system that solved the 50-year-old protein folding problem by predicting the 3D structure of over 200 million proteins, as proof of concept1

1

. The system, which earned Hassabis the Nobel Prize in Chemistry in 2024, now serves over 3 million researchers worldwide.

Drug Discovery and Isomorphic Labs

Hassabis is actively translating his scientific vision into commercial reality through Isomorphic Labs, an Alphabet subsidiary dedicated to "solving" disease1

1

. By shifting drug discovery from traditional "wet labs" to in silico computer simulation, he believes the process can become "1,000 times more efficient." The company has already advanced to pre-clinical trials for cancer drugs, with plans to reach clinical trials by year's end5

5

. Hassabis anticipates AI will significantly boost the productivity of scientists and experts, particularly in cross-disciplinary research that requires understanding multiple fields simultaneously3

3

. His excitement centers on models eventually serving as "co-scientists" as they grow more autonomous, capable not just of solving conjectures but formulating the right questions and hypotheses—a trait that separates great scientists from good ones.

AI Risks and Global Cooperation

Despite his optimism, Hassabis warns that urgent attention must focus on AI risks, particularly in biosecurity and cybersecurity3

3

. "The current systems are getting pretty good at cyber," he cautioned, stressing the need to ensure "defences are stronger than the offences." He divides risks into two categories: societal risks when bad actors misuse AI, and technical risks when systems behave in unexpected and potentially harmful ways2

2

. Addressing these threats requires scientific guardrails, robust monitoring systems, and global cooperation. "In order to mitigate some of the risks, we're going to need international collaboration," Hassabis stated, emphasizing that responsibility extends beyond technologists to include governments, scientists, artists, and philosophers4

4

. His call for careful navigation reflects the magnitude of transformation ahead: "We've got to try and navigate this moment very carefully."

Google's Innovator's Dilemma and Internal Transformation

For Google, the path to this renaissance presents what Hassabis calls a "classic innovator's dilemma"1

1

. The rise of generative AI represents an existential pivot for the $3.9 trillion tech giant, forcing it to risk disrupting its core search business. "If we don't disrupt ourselves, someone else will," Hassabis declared, a philosophy that drove the 2023 merger of Google Brain and DeepMind into a unified research entity under his leadership. The consolidation was necessary to pool the "enormous compute power" required to train frontier models like Gemini. "Bringing the two groups together and trying to combine the best of both cultures has been great," Hassabis said, likening the combined entity to a "nuclear power plant that's plugged into the rest of this amazing company." The strategy appears effective: following releases of models including Gemini 3 and the viral image generator "Nano Banana," Alphabet shares soared approximately 65% by year's end. Hassabis maintains the company has "crossed the watershed moment" where AI models can act as useful assistants in high-level research, positioning Google DeepMind against rivals like OpenAI in the race to unlock advanced AI's full potential.