Most scientists spend a lifetime pursuing a single breakthrough.

Demis Hassabis used one breakthrough to prove he was ready for the next one.

After helping create AlphaFold—a system that predicted the three-dimensional structures of more than 200 million proteins and transformed biological research—Hassabis shared the 2024 Nobel Prize in Chemistry. But protein folding was never his ultimate destination.

His long-term objective has remained remarkably consistent for more than three decades: understand intelligence well enough to build it.

Today, as CEO of Google DeepMind⁠, Hassabis is leading one of the world’s most ambitious efforts to develop Artificial General Intelligence (AGI)—systems capable of learning, reasoning, planning, adapting, and solving problems across many domains rather than performing a single specialized task.

To understand why he may be uniquely positioned to pursue that goal, it helps to understand the unusual path that brought him here.

Hassabis first gained recognition as a chess prodigy. By age 13 he had reached master level competition. While many young chess players focus on winning games, Hassabis became fascinated by a deeper question: what mental processes allow humans to think strategically, imagine future possibilities, and make decisions under uncertainty?

That curiosity led him into computer programming.

As a teenager he became a lead developer on the groundbreaking video game Theme Park. Creating virtual worlds taught him something chess could not: how to build complex systems, simulate environments, and create agents that interact with those environments.

Most successful game developers would have stayed in the industry.

Hassabis left.

He earned a degree in computer science from University of Cambridge and then made an even more surprising move. Instead of pursuing a lucrative technology career, he returned to academia to study the human brain.

At University College London he investigated how the hippocampus enables memory, imagination, and the ability to mentally construct future scenarios. His research helped demonstrate that memory is not simply a storage system. The brain actively builds internal models of reality, allowing humans to imagine events that have never happened and plan for situations that do not yet exist.

That insight would become one of the foundational ideas behind modern AI.

In 2010, Hassabis co-founded DeepMind with a mission that sounded audacious at the time:

Solve intelligence. Then use it to solve everything else.

The company’s early breakthroughs came through games. DeepMind systems learned to play Atari games, mastered Go, and defeated world champion Lee Sedol in 2016. Many observers thought the company was building game-playing machines.

Hassabis saw games differently.

Games provided controlled environments where learning systems could be trained, tested, measured, and improved. They were laboratories for developing general-purpose learning algorithms.

AlphaFold became the first major demonstration that those algorithms could move beyond games and solve a real scientific problem.

The success was historic.

What once required years of laboratory work and millions of dollars could often be predicted computationally in hours. Researchers around the world suddenly had access to protein structures that had never been experimentally determined.

The Nobel Prize recognized that achievement.

But for Hassabis, AlphaFold was also proof of something larger.

It demonstrated that AI could accelerate scientific discovery itself.

Now the focus has shifted to what comes next.

Rather than building systems that merely answer questions, DeepMind is working toward AI agents capable of reasoning, planning, conducting research, using tools, interacting with the physical world, and generating new scientific knowledge. Hassabis frequently describes future AI systems as collaborators that could help solve challenges in medicine, energy, materials science, mathematics, and climate research.

The vision extends far beyond chatbots.

Imagine an AI scientist capable of reading every published paper in a field, designing experiments, identifying overlooked connections, proposing new hypotheses, and collaborating with human researchers. Imagine digital systems that can reason across biology, chemistry, physics, engineering, and medicine simultaneously.

That is the direction DeepMind is heading.

Whether AGI arrives in five years or twenty remains uncertain. Many technical challenges remain, including reasoning reliability, long-term planning, memory, safety, alignment, and scientific validation.

But if history is any guide, Hassabis has been preparing for this challenge his entire life.

Chess taught him strategic reasoning.

Video games taught him simulation and engineering.

Neuroscience taught him how biological intelligence constructs models of reality.

DeepMind gave him a platform to test those ideas.

AlphaFold proved that the approach could solve one of science’s most important problems.

The Nobel Prize was not the finish line.

For Demis Hassabis, it may have been the strongest evidence yet that the search for artificial general intelligence is on the right path.