DeepMind created an AI that took three days of self-learning to become the Chinese game's greatest ever strategist.

By Joe McDonough

Posted on October 20, 2017

An Artificial Intelligence (AI) program has in a matter of days developed skills that have taken humans thousands of years to master.

Google’s AlphaGo Zero, took just three days to become unbeatable in the Chinese strategy game of Go.

Like the unpopular kid in class, it will learn simply by playing alone, and against itself.

Better, it has since discovered combinations of moves that human Go masters had never even conceived.

This isn’t the first time Google’s DeepMind lab in London has created an AI program superior to human Go players. But it is the first time the program has done so after learning the game from scratch.

Last year, AlphaGo famously triumphed over South Korean Go grandmaster Lee Sedol, winning a head-to-head series 4-1.

But that program developed its winning techniques after being exposed to thousands of amateur and professional games. ‘Zero’ as Engadget puts it, “foregoes the need for human insight altogether”.

“Like the unpopular kid in class, it will learn simply by playing alone, and against itself.”

How the AI program learns

DeepMind, a subsidiary of Google’s parent company Alphabet, explained the process in a blog post.

“The system starts off with a neural network that knows nothing about the game of Go. It then plays games against itself, by combining this neural network with a powerful search algorithm. As it plays, the neural network is tuned and updated to predict moves, as well as the eventual winner of the games,” the company wrote yesterday.

“This updated neural network is then recombined with the search algorithm to create a new, stronger version of AlphaGo Zero, and the process begins again. In each iteration, the performance of the system improves by a small amount, and the quality of the self-play games increases, leading to more and more accurate neural networks and ever stronger versions of AlphaGo Zero.

“This technique is more powerful than previous versions of AlphaGo because it is no longer constrained by the limits of human knowledge. Instead, it is able to learn tabula rasa from the strongest player in the world: AlphaGo itself.”

Zero proved just how far AI technology has come. After just three days of machine learning, it trounced the earlier version that outplayed grandmaster Sedol 100 games to nil.

DeepMind revealed it is looking into the best ways to utilise its groundbreaking AI.

“If similar techniques can be applied to other structured problems, such as protein folding, reducing energy consumption or searching for revolutionary new materials, the resulting breakthroughs have the potential to positively impact society.”