Score One for Humanity: Google’s Go Machine Finally Loses to Korean Champ
The machine can be beaten after all. Today, at least.
Korean Go champion Lee Sedol finally bested Google’s computer program in the ancient game that scientists at companies like Google and Facebook have been using to challenge their machine learning technology. Google’s machine, AlphaGo, had beaten Sedol three straight times before Sunday when Sedol notched his first victory.
The AlphaGo program, created by Google’s DeepMind division, includes much of the same technology the search giant uses for other products with human-like characteristics, such as photo recognition software and virtual assistant technology. Beating a human Go champion proves the company’s tech is very, very advanced but losing to a human go champion proves there’s still room for improvement.
Just how human is AlphaGo? DeepMind co-founder Demis Hassabis tweeted during the match that AlphaGo got “confused” and didn’t realize it had made a mistake until almost 10 moves later. Just like a human, the machine is expected to learn from its mistakes.
Mistake was on move 79, but #AlphaGo only came to that realisation on around move 87 間違いは79手目。しかし AlphaGoは87手目にやっと気付いた。
Mistake was on move 79, but #AlphaGo only came to that realisation on around move 87— Demis Hassabis (@demishassabis) 2016年3月13日
People around the world have been following this man v. machine battle closely. Lee Sedol was trending on Twitter in multiple regions following the victory.
If you have six hours of extra time today, you can watch the match on YouTube.