Yesterday I watched The Thinking Game, a documentary by Google, more specifically by the DeepMind team, which I've followed for many years. Their projects have always been interesting, but this new documentary brings something that goes beyond impressive technology.

The beginning: brute force in chess

To understand what DeepMind is doing, it's worth recalling the context. When IBM created Deep Blue to play chess against Kasparov, the idea was simple: prove that a computer can beat the greatest player in the world. Chess was synonymous with intelligence, so a computer champion would, by extension, be an intelligent machine.

It worked. It was a milestone. The media covered it intensely, and the world was impressed. But a more precise analysis reveals that it was, essentially, computational brute force: hundreds of millions of positions per second, evaluating many moves ahead to decide the next one. The computer wasn't thinking — it was calculating faster than any human could.

Today we have engines like Stockfish that simply destroy the best players in the world. But the method is the same.

Go: when brute force isn't enough

Go is different. The board is larger, the pieces are just black and white, and the number of possible games is greater than the number of atoms in the observable universe. No computer, no matter how fast, can simply calculate everything.

To win at Go, DeepMind needed a different approach: artificial neural networks, the Deep Neural Networks. Instead of analyzing all possibilities, the model is trained on an enormous volume of games and learns to recognize patterns, to generalize. It doesn't need to see everything: it learns to estimate what is probably best and follows that path.

The result was AlphaGo, which defeated one of the world's greatest Go players, South Korean Lee Sedol. And that mattered in a way that goes beyond technology.

Why AlphaGo was more than a victory

Go is part of East Asian culture in a profound way. In China, it's included among the "Four Arts" of the classical scholar, alongside music, calligraphy, and painting. In Korea and Japan, it carries similar cultural weight. In all these countries, mastery of Go has always been associated with intelligence and human capability. Unlike chess, where defeat to a machine was already expected after Deep Blue, the fall in Go was visceral. People were moved. There was a psychological, philosophical, and cultural dimension to that moment.

DeepMind's previous documentary, about AlphaGo, captures this very well. It is, in my opinion, more emotionally moving than The Thinking Game, but The Thinking Game brings something far more important.

There is something greater happening when AI begins to question what makes us special. We've been through this before: when humanity discovered that Earth was not the center of the solar system, nor of the galaxy, nor of the universe, it profoundly shook the perception we had of ourselves. Questioning the exclusivity of human intelligence is one more step in that direction — a lesson in humility we are still learning to process.

AlphaFold and the greatest achievement of AI to date

The Thinking Game documents AlphaFold, which I consider the greatest practical achievement of artificial intelligence so far.

For 50 years, scientists around the world invested time, money, and study trying to solve the protein folding problem: how does a sequence of amino acids transform into a specific three-dimensional structure? It was an open scientific problem, a collective effort without a solution in sight.

Humanity could discover a few proteins per year, at great cost and effort. AlphaFold not only solved the problem, but went further: it mapped and made available 200 million proteins to the entire scientific community. It simply ran and discovered all possible patterns in nature.

For that, the work earned the Nobel Prize in Chemistry. It's deserved.

The difference between interpolating and discovering

The question people always ask about AI is: is it just interpolating training data, or is it capable of solving genuinely new problems?

It's not the same as training a model on a quiz with questions and answers and then asking about the same quiz. When AI manages to solve a problem that no one has solved before — like protein folding — the conversation changes completely.

And the doors AlphaFold opened are enormous: drug development, solutions for diseases, possible answers to global warming, even ways to destroy the plastic we've accumulated on the planet. These are concrete possibilities, not abstractions.

Conclusion: hope with feet on the ground

The Thinking Game is an important documentary for understanding the moment we're living in. It's hopeful, perhaps overly hopeful at times, with a sense that something absurd and transformative is about to happen. That part is more abstract and without many solid conclusions.

The question of general AI — whether we're close or far, whether it will happen in 30 years or a thousand years or never — remains without an answer. It's too complicated for any honest prediction.

But AlphaFold is real. It's useful. It's new. It did something that science couldn't do on its own, and DeepMind had the generosity to release everything to the world.

That's what matters most in the documentary.