Google's AI Remains Undefeated
AlphaGo is a "narrow AI" developed by Google's DeepMind to specifically play the ancient game "Go." In October 2015 it made history for the first time by being the first and only computer program to beat a human at the game Go. Go is an "abstract strategy board game" with two players and is roughly 2,500 years old. Why pick Go? Well, despite its seemingly simple appearance it is more complicated than chess and has more possibilities than the total number of atoms in the universe. So, you can see why it is the perfect test for an AI.
Well, AlphaGo recently made history again! Just last week, May 2017, AlphaGo won its second game against a human - Ke Jie, a 19 year-old Chinese man. He lost a three game match with AlphaGo. According to the developers at DeepMind, who watched the match with great scrutiny, Ke Jie came the closest anybody has to beating AlphaGo. In fact, his first 50 or so moves were almost perfect!
Demis Hassabis went as far as tweeting "What an amazing and complex game! Ke Jie pushed AlphaGo right to the limit."
Ke narrowly lost the first match and was even a little shaken up about the loss. However, by the second game onlookers noticed he had loosened up a bit, although he still seemed wary of the machine which he described as "godlike" in its game playing abilities.
AlphaGo, with this victory, continues its legacy as the best Go player in the world. No one to challenge it, even the champions of the game, have been able to beat it.
This isn't the end of testing AlphaGo, though. Later on it will be tested as part of a human team against two other Go players, having to work in tandem with its human teammate. This will be done in an effort to teach AlphaGo how to better work with humans, as well as understand their decision making in the realm of Go.
What does this mean for the future of AI? Clearly, all this time, money, and news coverage aren't just for technology that is better than humans at board games. It is important to note how powerful AlphaGo is. In fact many in the AI industry believed that a "Go program as powerful as AlphaGo was at least five years away." This shows us that AI technology is progressing at a faster rate than previously thought possible. The development, programming, and team behind AlphaGo show just how intensive and powerful the field of AI is.
It's also important to understand that while this is impressive, AlphaGo can only play Go and does not have general purpose intelligence. It doesn't have common sense or autonomy or, well, will. AlphaGo is completely under human control and has no capabilities of defecting to a different purpose. So, there is no need to be scared yet...just remember to stay informed.
The above image was taken by wikimedia user Donarreiskoffer and is liscensed Creative Commons 3.0