Lee Sedol, 17-time world Go Champion, studied the 36 black and white stones sitting on the board, waiting for his opponent's next move. Fans around the world waited with him, watching the two Go players locked in a historic battle.
Go is an ancient and famously complex board game, so complex it has more possible configurations than atoms in the observable universe – two times more. After the first two moves in Chess, there are 400 possible next moves. In Go, that number is 130,000.
When Sedol's opponent placed the next stone, everyone was shocked. The move, eventually known as "Move 37," looked strange, even ill-advised, to the experts. However, as the game progressed, it became clear the move was brilliant—the beginning of Sedol's defeat.
A year earlier, another expert Go player, Fan Hui, played and lost to the same opponent. The "player was a little strange, but a very strong player." Hui commented, "he can show humans something we never discovered." Like Hui, other commentators referred to that Go player as a "he" when, in fact, it was a machine called AlphaGo.
Lee Sedol would win only one of five games against AlphaGo. Humans around the world celebrated his one win in Game 4. Even the creators of AlphaGo were overjoyed to see Sedol win. Speaking after the game, Sedol, with his penetrating yet gentle style, observed, "I heard people shouting in joy, and I think it is clear why. It seems like we humans are so weak and fragile, and this victory means we could still hold our own."
We, humans, have welcomed smart machines into our homes. They respond to our voices and learn the rhythms of our daily routines. We talk to them, calling them by names like Alexa, Siri, and AlphaGo. In turn, they are challenging us to consider what makes us uniquely human.
AlphaGo took more than three years to build using an approach called supervised learning. The program's creators gave everything that humans knew about Go to its sophisticated algorithms, which then used computing power to beat the best minds humans had to offer.
Its successor, AlphaZero, learned in a vastly different way called reinforcement learning. In this zero-knowledge approach, the creators gave it only the basic rules of the game before cutting it loose. When AlphaZero took over, it played games against itself, became both the student and the teacher, and invented strategies unconstrained by the limits of human knowledge.
AlphaGo beat Lee Sedol after three years of assisted learning.
AlphaZero beat AlphaGo after 72 hours of independent learning.
Since then, AlphaZero has applied its ability to learn unsupervised to other applications, like quantum computing, solving tricky problems to the surprise of human experts. Expert knowledge. That has always been a specialty of humans, and we’ve used it to define and shape our world. What, then, will be our advantage when machines create and use them better than we?
Wisdom.
That is a very tricky concept to define when we live in a vast reality with no detailed instruction manual. Wisdom is more closely aligned with purpose and meaning, and requires you to decide, who you are, where you are, and what you value. These decisions can not be answered by expert knowledge nor by incredible computing power. But the machines in our future, along with their growing knowledge, will usher in a new age where wise decision will be imperative. Which is why, in the years before his death, Stephan Hawking gave humanity this advice. "Our future is a race between the growing power of our technology and the wisdom with which we use it. Let's make sure that wisdom wins."
Therefore, I encourage you to cultivating wisdom, which I believe begins in wonder. Wendell Berry has some wisdom for us here.
"We must learn to acknowledge that the creation is full of mystery; we will never entirely understand it. We must abandon arrogance and stand in awe…. I do not doubt that it is only on the condition of humility and reverence before the world that our species will be able to remain in it."
Comments