If you are interested in artificial intelligence and want a glimpse of what the AI-driven future might look like, you may want to watch the just-released AlphaGo movie on Netflix. The movie chronicles the AlphaGo technology winning four-out-of-five matches against the world’s greatest Go player (Lee Sedol).
The maturation of deep learning and neural networks are a couple of the more important breakthroughs in the field of artificial intelligence. The AlphaGo project was undertaken by Deep Mind to see how well their neural network algorithm utilizing deep learning could compete against humans at the strategy game called Go. The game of Go is so complicated that it can’t be beat via brute force computing even with today’s supercomputers. The AlphaGo technology needed to learn how to play the game.
While impressive that the AlphaGo technology was able to defeat the top-ranked human players, it was how the technology accomplished this feat that was the most impressive. The technology learned via interactions with humans and then built upon this experience to play in ways different from the way humans historically approached the game. There are a couple points in the movie in which the match commentators allude to AlphaGo making mistakes only to realize at a later point in the game that the technology was applying a winning strategy. It was a humbling experience for the players, commentators, and fans.
Within a year of AlphaGo defeating the top-ranked human Go player, Deep Mind released the next generation of the technology —AlphaGo Zero. The new technology learned to play Go without human interaction. It learned how to play the game on its own and it soundly defeated the original AlphaGo.
These deep learning systems are rapidly evolving and will continue to become increasingly sophisticated and human-like. Similar to what happened with the Go matches, the technology may help broaden our perspective and help us think of new ways to approach solving complex challenges.