Lee Se-dol Versus AlphaGo – AI Comes of Age

Go game boardThe tech world is buzzing at the moment – the reason being a simple analogue game played that is being played out between one man and a computer. Lee Se-dol got off to a very shaky start, losing three games in a row, but he won the fourth, proving that AlphaGo is certainly not unbeatable – at least, not yet.

Lee Se-dol, a pro Go player, is playing against the latest AI machine, AlphaGo. As the name suggests, this machine has been built with one goal in mind – to win at Go. Unlike the chess playing supercomputers of the past, such as Deep Blue, AlphaGo has been programmed to be able to learn and apply intelligence. Also, Go is considered to be a much more complex game than chess, with a huge number of variations and strategies.

AlphaGo was built by a team at Google, a company that is starting to put AI at the forefront of its services. From driver-less cars to search engine that can learn, AI is starting to steer new technology. Google’s AlphaGo system was developed by a British company, formally called DeepMind, before Google acquired it two years ago.

AlphaGo Strengthened by Loss

While Lee Se-dol was very pleased with his victory, saying, “I’ve never been congratulated so much because I’ve won one game”, the Google team is confident that the defeat will simply allow AlphaGo to become an even better player.

The first three games all went AlphaGo’s way, but only by a small margin. In fact, Lee was ahead for much of the games, only losing out during the final stages. Lee descrobed AlphaGo as playing a “nearly perfect game”.

AlphaGo, like a human player, is able to learn from its mistakes. It will only likely fall into one trap once – unlike a human player, who could potentially stumble into the same trap several times before learning how to defend and counter-attack successfully.

Why Go?

Go is thought to be one of the most complex board games in the world. It was developed thousands of years ago in China, and although it looks very simple, it is a highly complex game. Many human players still rely on instinctive play – although, it is hard to say how instinctive their moves are, as they may be drawing from experience and making rapid calculations at a subconscious level.

If you fancy a game yourself, head over to Go Game Guru where you can learn what go is and start learning how to play go, among other things. There is also a pretty good discussion on the Se-dol – AlphaGo game. The Wikipedia page is also gives a good intro to the game.

Related posts