On the morning of March 9th at the Four Seasons Hotel in Seoul, South Korea, the world changed forever. A computer program called AlphaGo used the relatively new social science of cognitive computing to strategize and beat one of the top masters in the world at the ancient Chinese board game Go. It was hailed as one of the most defining moments in the development of artificial intelligence so far.
At NETSCOUT, we watched the development of the AlphaGo program, originally created by British company DeepMind before being acquired by Google in 2014, with great interest. Cognitive computing is something we do every day at NETSCOUT to ensure that our solutions deliver the digital strategies of our clients with maximum efficiency. We like to sometimes compare ourselves to conductors helping to keep the trains of digital networks running on time, and while that is an accurate description of our service assurance platform, it doesn’t really dive deeply into the intelligence that our products provide. And like with AlphaGo and the science of cognitive computing, you can expect a lot more advances from us in the near future.
Since cognitive computing is so important to us, and in our minds to the digital transformation business need to undergo to be successful, we are devoting a three-part blog to this disruptive technology. First we will write about cognitive computing in general, and AlphaGo’s pivotal role within this practice. Then we will talk about examples where it can be used to vastly improve operations in various markets. And finally, what NETSCOUT is planning to do that will transform businesses and the CIO role.
Must Pass Go
Let’s start by addressing the misconception that AlphaGo’s victory is “no big deal” because IBM’s Deep Blue supercomputer already beat chess grandmaster Garry Kasparov in 1997. Deep Blue was not an example of cognitive computing. Instead, Deep Blue’s victory is an example of using brute force computing to defeat a human counterpart.
At a tactical level, players in chess have about 20 possible moves each turn, while in Go it’s about 200. That leads to many more possible game variations. According to the website chess.stackexchange.com which tracks all things chess, by the second move in a chess game, there are 72.084 possible game combinations. By the third, it’s nine million and by the fourth it jumps to 318 million. That’s a lot, but not out of reach for a dedicated supercomputer to process in a matter of seconds. The Go game by contrast has an almost unlimited number of possible games, something akin to the number of atoms in the universe according to DeepMind officials.
Go is played on a 19 by 19 square grid with two players alternatively placing white and black tiles. Players can capture their opponent’s tiles by surrounding them. It’s almost impossible to tell who is actually winning until the very end of the game, and many of the top masters rely on instinct as opposed to any direct plans one might learn with other games. To win, AlphaGo had to rely on superior pattern recognition and actual learned strategy, just like a human player. Top Go player Lee Se-dol, who was beaten three games to zero in the best of five series, was said to be highly impressed with the program, like he was playing an actual human.
What AlphaGo proved, beyond just being able to play a game, is that cognitive computing is here, and it works. The reason people use computers to examine data is because computers are faster than humans. But what they lack is the contextual understanding to make sense of things, to connect the dots and mirror an actual human way of thinking. AlphaGo was able to do that.
And that is what has everyone here at NETSCOUT so excited. By adding metadata to the data streams that we are already monitoring, we can then use cognitive computing to help the computers decide how to react within context. This can lead to self-healing networks and computers that fix themselves on the service assurance side of our business, which is a vital baseline for any organization. But that is only the beginning. What it can also do is allow for limitless innovation, re-programming and re-routing of networks in unique and unforeseen ways for maximum efficiency. It also can help improve or even perfect the customer experience, giving them exactly what they need faster and, dare we say, more efficiently than even a human.
Stay tuned for part two of this series, when we will dive into some specific examples of how cognitive computing can be a real game changer for a variety of businesses, sectors and vertical markets. We are in for an exciting future, and it’s going to be here sooner than most of us expected.