AlphaGo

What is AlphaGo?
AlphaGo is a tight AI, a computer program developed by Google DeepMind for playing Go, a Chinese strategy game for two chess-like players. AlphaGo is the first AI program to beat a professional human player, Fan Fan Hui, on a large board with no handicap in October 2015. Then in March 2016, he beat one of the world's top ranked human players, 9-dan Lee Sedol, and won four out of five games.

The AlphaGo project was started in 2014 as a test bed to see how well Google DeepMind's neural network algorithm using in-depth learning could rival Go. The AlphaGo algorithm is a comb in ation of tree searching and machine learning techniques and reinforced with extensive training with humans and other computer players. It uses Monte Carlo tree search and is guided by a policy and values network implemented using deep neural network technologies. The policy network is trained and helps the AI predict the next move most likely to win, while the advertising network is trained to narrow the search tree and determine the value of those positions, estimating the winners in each position and not until the end of the game .

AlphaGo was first fed historical plays by human gamers and used a database of around 30 million moves to mimic human games. Once the AI reached a certain level of proficiency, it was further trained by playing against its own instances and using reinforcement learning to improve and learn more.

In October 2015, a distributed computer version of AlphaGo played and defeated Fan Hui, a 2-dan European Go Champion. It was the first time a computer program beat a professional gamer on Go. Fan Hui then helped out as an advisor to the DeepMind team months after his defeat. In March 2016, AlphaGo competed against Lee Sedol, one of the highest ranked players in the world, after reaching the top level of 9-Dan. He won four games for Lee which was a big breakthrough in AI research as the DeepMind algorithm used by DeepMind could be used for deep learning and neural networks for other purposes as it was not actually programmed to go, but taught became how to play Go, opening up a whole new world for AI research.