Go is a traditional Chinese strategy board game that is approximately 2,500 years old, and the current grandmaster of Go just got beat by a computer. Go is a really difficult game. Each possible move has around 250 options, so regular computers can’t learn to play the way they can play chess, which usually only has about 35 options for moves. Google was putting it’s artificial intelligence, called AlphaGo, to the test against an unnamed grandmaster of the game, and the results were stunning. Many are hailing this win as a big moment for AI, however, one analyst Nick Bostrom isn’t convinced. He believes the results of this historic game shouldn’t be at all surprising. Many also think that before we delve into creating artificial intelligence, we need to reflect on what can happen with this amazing and yet also frightening technology.
AlphaGo uses not only deep learning to become intelligent, but it also uses reinforcement learning. Deep learning is the type of technology that allows Google Photos and Facebook pics to recognize certain people, and what allows Skype translations to be so efficient. Reinforcement learning is kind of like sending a robot to a casino, and is most effective in gaming scenarios. The AI systems play games against each other and evaluate which moves have the highest returns. It then remembers those moves and levels up their playing skill, both a frightening and exciting prospect.
Facebook and Google have been in a bit of a “Go” competition, with FB also building an AI system to learn how to play the game. Before, researchers thought it might be kind of a while – maybe as long as ten years – before AI could defeat a human player such as a grandmaster, but leave it to Google to prove them wrong, and win the race.
There are no upcoming events.
© 2015 All DigitalApex. All Rights Reserved.