Google’s AlphaGo AI on Tuesday shocked the AI community all over the world by beating world’s number 1 “Go” player, Ke Jie. Developed by Google’s DeepMind Technologies, AlphaGo is an AI program specialising in the Chinese board game “Go”.
AlphaGo’s win is widely being considered a landmark moment in AI development. However, the writing is not on the wall yet, as AlphaGo is set to face off against Jie two more times as part of a 3 match series. As it stands, the advantage clearly lies with the machine rather than man.
This defeat of a world champion by a machine in a game of wits is not a first-time occurrence. We can look back to the now famous defeat of the world chess champion Garry Kasparov by IBM’s Deep Blue AI in 1997 as an example. However, AlphaGo’s win is much more significant. This is particularly due to the nuances of the game of Go itself.
The 2,500-year-old Chinese game Go is often considered one of the bigger milestones in AI development. In fact, there is an entire field of AI, called “Computer Go“, that is dedicated to developing AI capable of beating human Go players. The reason AI researchers are so obsessed with Go is that it requires intuition, creative and strategic thinking. This makes Go a much more difficult proposition than Chess for AI.
AlphaGo has proved that this is not a one-off either. A year ago, AlphaGo also managed to beat Go genius Lee Se-dol 4-1 in a 5 game series. AlphaGo’s tremendous successes prove that machine learning is the way forward in AI development. It uses two sets of “deep neural networks”, with millions of connections that mimic the structure of a human brain. After initial programming, AlphaGo played millions of games with itself to learn the nuances of the game.
All of this points towards a conclusion that many already suspect – we may be at the dawn of the golden age of AI. Deepmind’s accomplishments show just how far AI development has come. It seems within the realm of possibility to imagine a near future where AI controls our cars, homes and even our cities.
It seems Google may very well usher the AI revolution sooner than we think. Their commitment to the technology is apparent from their decision to open up their TensorFlow deep learning framework. They are attracting small developers and individuals to their AI ecosystem by providing an impressive array of machine learning tools. By doing this, they are trying to stamp their influence on the upcoming AI ecosystem which is surely going to dominate the world soon. In many ways, this strategy is similar to how Google won the smartphone OS war. If current developments are any indication, they may just replicate this success with AI too.