I remember when I was eight years old and my parents were watching Matrix on television. Truthfully, I didn’t understand much, neither the story nor the necessity of violence, but I do remember vaguely the battle between humans and “computers”. And here I am twelve years later, witnessing the battle of wits between humans and artificial intelligence.
It was about 20 years ago when artificial intelligence made it on the headlines and proved its superiority over humans in complex decision making games. On May 11, 1997, Deep Blue, an IBM computer, challenged and championed over Garry Kasparov after two wins, one loss, and three ties. Though the impact was already clear with people worldwide talking about the potential dominance of artificial intelligence in the coming years, many were still pessimistic, as they thought chess was simply a game and not much more. Today, the world is witnessing another historical clash: Google’s AlphaGo versus Lee Sedol, a South Korean professional Go player of 9-dan rank.
Go, originating from ancient China more than two centuries ago, is not only one of the oldest board games played today, but also the most strategic and complex game that exists today. For instance, the number of possible games is incredibly vast, spanning to a total of 10761, compared to the estimation of 10120 possible games in chess. Due to this discrepancy in complexity and number of strategies, even when Deep Blue beat Kasparov, people were adamant that artificial intelligence could not beat humans in a game of Go, in which each move would provide drastically different outcomes. Even Demis Hassabis, the founder of Deep Blue, acknowledged the greater possibility of different moves in Go over chess, stating “Go gives the player about ten times as many options, resulting in a massively higher branching factor that is far harder for any AI to deal with.”
Last October, however, the world was taken aback when AlphaGo beat Fan Hui, a professional Go player with experience in coaching the French national team. Powered by two deep neutral networks that effortlessly narrow its decisions by shortening the tree diagram of possibilities, AlphaGo relies on the most advanced technology yet to make the best possible decisions in this ancient board game. However, rather than relying solely on its originally programmed algorithms through data from historical real-world matches, AlphaGo goes as far as to constantly play against itself, reinventing its own decision-making processes.
In a five-game match against Lee Sedol lasting from March 9, 2016 to March 15, 2016, AlphaGo won a total of four– losing only once in the fourth round. Its victory in the first round dazzled not only Go enthusiasts but even those completely new to the game who had come to the realization that an AI had beaten humans at a game with more possible moves and positions than the number of atoms in the observable universe. However, the one victory that Lee Sedol achieved in the fourth round was enough to signify that the AI wasn’t always right at the very least.
A clash against artificial intelligence with a grand prize of a million dollars. Everyone from Go players to newcomers tuned in everyday to witness history in the making. But what are we missing? Why is Google hosting this match anyways? Though AlphaGo did lose one round, the sign of its superiority over Lee Sedol was enough to carve in people’s minds the extent to which AI can perform tasks better than humans. Could Google be making a bold statement in the near future in regards to AI taking over most people’s jobs? Sure, it may be too early to assume, but we’ll see as Google prepares another encounter against humans in another popular game - Starcraft.