Google’s AI "DeepMind" Trounced Human Mind - The Last Day of Go

By March 14, 2016 at 8:44 am

It happened - it happened ten years too early. The supposedly world's strongest master of the game of Go bowed to an artificial intelligence "in the person of " AlphaGo consecutively losing the first three of the scheduled five matches, March 9, 2016. Two more to go, but this "person" is likely to sweep both - the way he has beaten the living Go master three in a row.

Hello, I'm your navigator Nathan Shiga to pass on to you probably an episode of the century. It's no exaggeration, folks, because everybody says so, everybody in the world of Go laments March 9, 2016 marked the last of Go.

How did it happen? Why so soon?

Associate professor Ryutaro Ichinose of the National Institute of Informatics bares DeepMind had adopted a shocking approach in outplaying Lee Sedol:

"Yes, the Google AI has obviously learned an entirely different approach to the game of Go. The so-called deep-learning enabled 'him' to grope for winning patterns. Unlike conventional AIs, AlphaGo (DeepMind) learns from past failures and works out alternative moves learned therefrom.

"In other words, AlphaGo studies the situation and works out moves that he believes winning, instead of just making strong moves."

Ichinose explains AlphaGo must have an advanced neural network that effectively combines policy network and value network to think like a human to work out every move at each given situation.

President Ryo Shimuzu of UEI, a top-notch programmer and expert on AI, points out:

"AlphaGo's neural network helps learn things at its will, so to speak, as living organisms grow on their own."

What does all this amount to?

A think-tank EY Japan predicts the domestic AI market will expand to worth 87 trillion in 30 years. Last year it was mere 3.7 trillion - meaning the market will blow up 20 times in 15 years.

Folks, AI beating a Go master is not a big deal after all. Something is happening in the world of "intelligence".

Related News

Trending News