Who (or What) Are We Rooting For?

"Who am I rooting for?" I found myself asking as I watched the tension unfold on screen. The AlphaGo documentary had me on the edge of my seat, my emotions swinging wildly between excitement and unease. On one side was Lee Sedol, the human champion, embodying centuries of Go tradition and intuition. On the other, the AI system AlphaGo, representing the cutting edge of artificial intelligence. But behind AlphaGo stood Demis Hassabis and his team at DeepMind, their human hopes and fears as palpable as Lee's.

This wasn't just a game anymore. It was a microcosm of our evolving relationship with technology, our aspirations, our anxieties, and the blurring lines between human and artificial intelligence. As I watched, I realized I was rooting for both Lee and Demis, for human creativity and technological progress. This internal conflict mirrored the larger questions swirling in my mind about AI's rapid advancement and our place in a world where machines can outthink us in ways we never imagined.

No this wasn’t in 2024 but in 2016, years before ChatGPT 3.5 felt like it changed everything in 2022. But that change was coming years ahead to those who were watching closely. 

The Minds Behind the Machine

Demis Hassabis is not just another tech entrepreneur. He's a prodigy whose journey to the forefront of AI began in childhood. A chess master by 13, Hassabis went on to design award-winning video games before diving into neuroscience and artificial intelligence. His background is a unique blend of game design, cognitive science, and computer engineering – a combination that would prove crucial in the development of AlphaGo.

Watching Hassabis in the documentary, I was struck by his quiet intensity and unwavering vision. For him, AlphaGo was never just about winning at Go. It was a step towards a much grander goal: developing AI systems that could think creatively and solve complex problems in ways humans couldn't. "I want to understand the big questions, the really big ones that you normally go into philosophy or physics if you're interested in," Hassabis once said. "I thought building AI would be the fastest route to answer some of those questions."

This vision set Hassabis and his team apart. They weren't just trying to build a better Go-playing machine; they were aiming to push the boundaries of what artificial intelligence could achieve. And in doing so, they hoped to unlock new insights into the nature of intelligence itself.

The Human Champion

On the other side of the board sat Lee Sedol, a titan in the world of Go. Lee wasn't just playing for himself; he carried the weight of human achievement on his shoulders. His journey through the match was a rollercoaster of emotions that resonated deeply with me and, I suspect, with anyone watching.

At the start, Lee exuded confidence. "I don't think it will be a very close match," he said before the games began. "I am confident that I can win." But as AlphaGo demonstrated its extraordinary capabilities, Lee's confidence gave way to shock, then despair. After losing the first three games, Lee appeared visibly shaken. "I am quite speechless," he said in a post-game interview. "I admit that it was a very clear loss on my part."

The toll of facing an opponent he couldn't fully understand was evident. Yet, in the face of this adversity, Lee showed remarkable resilience. He dug deep and managed to win the fourth game, achieving what many thought impossible. This victory, though not enough to win the overall match, was a powerful moment of human triumph.

The Move That Changed Everything

The pivotal moment came in Game 2, with AlphaGo's now-famous Move 37. It was a move so unexpected, so alien, that both Lee Sedol and the human commentators initially thought it was a mistake. But as the game progressed, it became clear that Move 37 was not just good – it was brilliant.

This moment encapsulated the promise and the challenge of AI. Here was a machine making a move that no human would have considered, demonstrating creativity in a domain we thought was uniquely human. It showed that AI could not only match human intelligence in certain areas but surpass it in ways we couldn't even comprehend.

I remember feeling a chill run down my spine as I watched this unfold. It was simultaneously thrilling and terrifying. On one hand, I was in awe of what Hassabis and his team had achieved. On the other, I couldn't help but wonder about the implications of an AI that could outthink the best human minds in such a complex domain.

The Human Element in a World of AI

Throughout the AlphaGo matches and beyond, what struck me most was the very human drama playing out alongside the technological achievements. The tension among the DeepMind team as they watched their creation perform was palpable. Their nervous energy, the mix of excitement and apprehension as they waited to see if years of work would pay off – it humanized the technological achievement in a way I hadn't expected.

Hassabis, despite his calm exterior, was clearly feeling the weight of the moment. In one interview, he reflected on the pressure: "It was pretty stressful... We've advanced the state-of-the-art in the field so that's fantastic, but we still a long way to go before we've solved it."

Lee Sedol's journey throughout the AlphaGo matches was a profoundly human story that resonated deeply with me. At the outset, his confidence was palpable – the assurance of a master at the peak of his craft, facing a challenger he believed he could overcome. But as the matches progressed, I watched with a mix of fascination and unease as that confidence slowly eroded, replaced by a growing realization that he was facing something beyond his comprehension.

The most haunting aspect was witnessing the gradual dimming of the light in Lee's eyes. It wasn't just about losing games; it was about confronting a fundamental shift in his understanding of the world. With each defeat, you could almost see the weight of this new reality settling on his shoulders. The man who entered the competition as the undisputed human champion was grappling with the idea that his mastery, honed over decades, could be surpassed by a machine in a fraction of that time. This wasn't just a professional setback; it felt like watching someone question their place in the universe. The quiet despair in his expression, the slump in his posture – these were the visible signs of an internal struggle that was both deeply personal and universally relatable. 

It struck me as a microcosm of the broader human experience in the face of rapid technological advancement: the mixture of awe, fear, and the need to redefine our understanding of our own capabilities.

From Games to Groundbreaking Science: AlphaFold

The story of AlphaGo doesn't end with its victory over Lee Sedol. Under Hassabis's leadership, DeepMind took the techniques developed for AlphaGo and applied them to one of biology's grand challenges: protein folding. This transition exemplifies Hassabis's broader vision for AI - using well-defined problems like games as a testbed to develop technologies that can then be applied to open-ended scientific challenges.

The result of this effort was AlphaFold, an AI system that can predict the 3D structure of proteins with unprecedented accuracy. To understand the significance of this achievement, we need to appreciate the complexity of the protein folding problem. Proteins are the fundamental building blocks of life, responsible for countless biological processes in our bodies. However, their function is largely determined by their three-dimensional shape, which in turn is determined by their amino acid sequence.

For decades, scientists have struggled to predict how a string of amino acids would fold into a functional protein. This challenge was so significant that it was often referred to as biology's "holy grail." Traditional experimental methods to determine protein structures, such as X-ray crystallography, were time-consuming and expensive, often taking years and millions of dollars for a single protein.

AlphaFold changed this landscape dramatically. By leveraging deep learning techniques and the vast amount of protein data accumulated over years, AlphaFold can now predict protein structures in a matter of days, sometimes hours. The accuracy of these predictions is remarkable - often matching or even exceeding the precision of experimental methods.

This breakthrough has profound implications for medicine, drug discovery, and our understanding of diseases. Many diseases are caused by misfolded proteins or involve interactions between proteins. With AlphaFold's capabilities, researchers can now better understand these processes and potentially design more effective treatments. For instance, AlphaFold has already been used to predict the structures of proteins associated with COVID-19, contributing to our understanding of the virus and potentially aiding in the development of treatments.

But perhaps even more impressive than the technology itself is DeepMind's approach to sharing it. In a move that surprised many, DeepMind made the results of AlphaFold's predictions for over 200 million proteins freely available to the scientific community. This decision has accelerated research across various fields of biology and medicine, with over 1.5 million researchers worldwide using AlphaFold in their work.

Watching this progression, from AlphaGo to AlphaFold, I was struck by the realization of just how far-reaching Hassabis's vision was. He and his team weren't just building game-playing AI; they were developing tools that could revolutionize scientific research and potentially save countless lives.

The success of AlphaFold also demonstrates the potential for AI to make genuine discoveries and generate new knowledge. Just as AlphaGo came up with novel strategies that influenced how human players approach the game, AlphaFold is predicting protein structures that would take human scientists years or decades to determine experimentally. This capability for genuine discovery is what makes AI so exciting as a tool for scientific advancement.

Looking forward, the technologies and approaches developed through projects like AlphaGo and AlphaFold are likely to be applied to an ever-wider range of scientific challenges. Hassabis envisions AI as a tool to explore what he calls the "tree of all knowledge," tackling root node problems that can unlock entire new branches of discovery. This could lead to breakthroughs in understanding the fundamental nature of reality, curing diseases, and even addressing global challenges like climate change.

However, as we celebrate these achievements, we must also grapple with their implications. The rapid advancement of AI raises important questions about the future role of human scientists and the potential risks of increasingly powerful AI systems. As we stand on the brink of new scientific frontiers, the promise of AI-driven discovery offers both excitement and challenges. The coming years will likely see AI playing an increasingly central role in pushing the boundaries of human knowledge, potentially unlocking secrets of nature and the universe that have long eluded us.

The Double-Edged Sword of Progress

As I watched these events unfold, I found myself grappling with the implications. The success of AlphaGo and AlphaFold raises intriguing questions about AI's future role in our lives. If AI can master Go and solve protein folding, what other complex problems might it tackle? Could it help us combat climate change, unravel the mysteries of dark matter, or develop new sustainable energy sources?

The potential is vast, but so too are the challenges. As AI systems become more powerful and autonomous, how do we ensure they remain aligned with human values and interests? The "Move 37" moment in the AlphaGo match showed us that AI can think in ways that are fundamentally different from human cognition. How do we prepare for a world where such thinking becomes commonplace across various fields of science and technology?

Hassabis himself is acutely aware of these challenges. "We've got to be careful," he has said. "We've got to really think through the ethics and how we solve these problems, and how we deploy these systems in the world." His approach to AI development, balancing ambition with caution, offers a model for how we might navigate these complex issues.

Embracing the Gray Areas

As I've reflected on the AlphaGo story and its aftermath, I've come to realize that there's no simple answer to the question of who we should root for. It's not a choice between humans and machines, or between progress and caution. Instead, we need to embrace a more nuanced perspective.

We can admire Lee Sedol's skill, creativity, and resilience while also marveling at the achievements of Hassabis and his team. We can be excited about the potential of AI to solve complex problems while remaining vigilant about its potential risks. We can push for rapid advancement in AI capabilities while also ensuring we have robust ethical frameworks and safety measures in place.

This nuanced view requires us to hold multiple, sometimes conflicting ideas in our minds simultaneously. It's not easy, but it's crucial if we want to navigate the AI-driven future responsibly.

Hassabis's approach embodies this nuanced thinking. He's pushing the boundaries of what's possible with AI, but he's also among the loudest voices calling for responsible development. "We need to use the downtime, when we have it, when the technology isn't fully formed yet, to solve a lot of these safety and ethical issues," he has said.

Charting Our Course in an AI-Augmented World

The story of AlphaGo isn't just about a machine learning to play an ancient board game. It's about us – our ingenuity, our fears, our hopes, and our capacity to adapt in the face of rapidly changing technology. 

It shows us the incredible potential of AI to push the boundaries of human knowledge and capability. But it also reminds us of the importance of our human qualities – our creativity, our resilience, our ability to learn and adapt.

What makes this documentary so compelling is how it weaves together multiple narratives. On one level, it's a thrilling contest between human and machine. But dig deeper, and you'll find it's also a story about the researchers behind the AI, their aspirations, and the implications of their work. It presents a nuanced view of technological progress, highlighting both its promise and its potential pitfalls.

As Demis Hassabis and his team at DeepMind have shown, the technology that can master a game like Go can also be applied to solve some of humanity's most pressing problems. The potential of AI to advance fields like medicine, climate science, and energy is enormous. But realizing this potential while mitigating risks requires careful thought, robust debate, and collective action.

The development of AI is not something that's happening to us – it's something we're actively shaping. Every decision we make, from the research we fund to the ethical guidelines we establish, plays a role in determining the future of this technology. Demis Hassabis and his team at DeepMind are at the forefront of this effort, but they're not alone. It's a collective endeavor that involves all of us.

In many ways I’ve committed my professional future to my own analysis of the future of AI. But that doesn’t mean I’m any more informed or clear than you are. Rather than trying to arrive at a definitive stance on AI, I've focused on staying informed, engaging in discussions with people from diverse backgrounds, and continually reassessing my views as new information comes to light. I've found inspiration in both Lee Sedol's resilience and Demis Hassabis's visionary yet cautious approach.

In the end, the question isn't just who we're rooting for, but what kind of future we want to create. The story of AlphaGo, of Lee Sedol and Demis Hassabis, reminds us that this future is in our hands. This is not the beginning or end – so it's up to us to write the next chapters of this extraordinary story.

I highly recommend watching the full 90 min documentary on YouTube.