From ‘Move 37’ to AGI: 10 years after AlphaGo, Google DeepMind CEO Demis Hassabis explains why AGI is the ‘ultimate tool’

“Move 37” is one of those rare moments in history where technology felt like art. Ten years ago in a hotel in Seoul, South Korea, a single click of a stone changed the course of computer science and propelled Artificial Intelligence (AI) into the global spotlight. Marking the decade anniversary, Google DeepMind CEO, Sir Demis Hassabis, has revisited that milestone, linking AlphaGo’s tech to the advancements of the last 10 years. He has explained why that particular moment was a definitive step toward achieving Artificial General Intelligence (AGI) – the ultimate tool to advance science, medicine, and productivity.In 2016, over 200 million people watched AlphaGo face world-champion Go player Lee Sae Dol in Seoul. During the second game of a historic five-match series, Google DeepMind’s AI system AlphaGo placed a black stone on the 37th line – a move so unconventional that commentators initially dismissed it as a mistake – but it wasn't and it proved to be decisive. About a hundred moves later, the stone was in exactly the right position for AlphaGo to win the game, said Hassabis.“It was a display of incredible foresight and the AI system’s ability to go beyond mimicking human experts and find entirely new strategies. The achievement heralded the beginning of what is now recognized as the modern era in artificial intelligence (AI). With a single creative play, the famous ‘Move 37,’ AlphaGo demonstrated the potential of AI and signaled that we now had the techniques to begin tackling real-world scientific problems,” Hassabis said in a blog.According to the Google DeepMind CEO, the fundamental architectures that powered AlphaGo, which is combining deep neural networks with reinforcement learning and advanced search, are now being used to build systems on the path to Artificial General Intelligence (AGI).Hassabis explained that AlphaGo first learned from games played by human experts, and then playing hundreds of thousands of games against itself, essentially improving itself and creating winning strategies – the very definition of AGI.“It was further proof of what I knew the moment we won the match in Seoul - the technology was ready to be applied to our real goal of accelerating scientific breakthroughs,” he said.The transition from the Go board to the laboratory is already yielding historic results. The ability to navigate a near-infinite “search space” – which allowed AlphaGo to master 10^170 possible board positions – was the same logic applied to solve the 50-year-old protein-folding problem. That project, known as AlphaFold, recently culminated in a Nobel Prize in Chemistry for Hassabis and colleague John Jumper.The legacy of Move 37 lives on in today’s most advanced models. Systems like Gemini now use “Deep Think” modes to solve complex mathematical proofs and coding challenges. It effectively uses the same “search and planning” DNA that surprised the world in 2016. However, Hassabis notes that the journey isn't over. “For an AI to be truly general, it needs to understand the physical world. We built Gemini to be multimodal from the beginning so it could understand not just language, but also audio, video, images and code to build a model of the world. To think and reason across these modalities, the latest Gemini models use some of the techniques we pioneered with AlphaGo and AlphaZero,” Hassabis noted.For Hassabis, the combination of Gemini’s world models, AlphaGo’s search and planning prowess, and specialised AI tool use will be critical for AGI. “True creativity is a key capability that such an AGI system would need to exhibit. Move 37 was a glimpse of AI’s potential to think outside the box, but true original invention will require something more. It would need to not only come up with a novel Go strategy, as AlphaGo impressively did, but actually invent a game as deep and elegant, and as worthy of study as Go,” Hassabis highlighted.