PC Threatens to Outsmart Man in Game of Go

This in incredible - simply unheard-of. It's something everybody had thought would be hard to come by. I'm Nathan Shiga and have here a thrilling news of AI nearly beating man's intelligence.

You see, I play a bit of Go, 5th dan, and can speak with some authority on what the game is about and what it takes to learn it. PC has beaten chess as DeepBlue dodged Kasparov in 1997 and four years ago downed a Japanese master Kunio Yonenaga for the first time to conquer the game of Shogi. Whereas chess has patterns as many as 120 times ten and shogi 220 times ten, the game of has over 360 times ten patterns to play and every white or black stone is of equal quality - unlike a pawn against a rook in chess.

The game of Go had long been regarded the last fortress for OC to challenge until an AI Go-playing PC developed by Goggle, AlphaGo, won five straight games against the current European champion Fan Hui of France last October. AlphaGo's feat was a shock not only to AI researchers across the world but to the world of the game of Go. It has happened to chess and shogi - what's so shocking about the same thing happening to Go?

It was a shock because a new technology in Ai science is at play this time - what is called "Deep Learning". In a nutshell, deep learning is a mechanism whereby PC is given its own brain to "perform" like man instead of discharging what is told. In deep learning, PC is given data on the problem to solve and and the answer to arrive at but not on how to solve it.

A Canadian university team succeeded in improving vision cognition accuracy using deep learning. AlphaGo is a happy spinoff of the technology. Next month, AlphaGo is challenging the current strongest Go master Lee Se-dol of South Korea.

This is something to watch, folks.