AI is escalating more swiftly than one could ever imagine. - Data Science Wale

Go has many more possible configurations than chess, so its computation is evidently difficult. But, yes, in this they have used a deep reinforcement learning approach, defines as to take many possible actions to get a reward and chooses an action through which it earns a best reward. There is another story of very old and popular game Mario, Sethbling a programmer developed a computer program who learned by itself how to play Super Mario World. That program, named MarI/O taught itself by doing different tries, example it learned from its own demise and tried to jump in every next try at each and every point it got killed previously. It followed a neural network approach to learn how to play game, this approach is same as human brain's working process. In 34 tries only it completed a whole level.