Google Inc (NASDAQ:GOOGL) is flexing its muscles in the innovation front having developed a machine according to the New York Times that can learn how to play and win video games. The new development marks a big leap in the development of computers capable of independent reasoning similar to IBM’s chess-winning deep blue computer.
Google Inc (NASDAQ:GOOGL)’s breakthrough on computers with independent reasoning could especially be of benefit on important things like voice recognition, search, and language translation. The technology according to researchers could be of most importance to robots and driverless cars. The program reportedly taught itself rules of over 49 Atari 2600 computer games developed in the 1980’S consequently figuring out the best strategy for winning.
The machine according researchers mastered various navigation, actions as well as positive outcomes then used them to improve outcomes. Tested on 43 games such as Space Invaders and Breakout, the machine outperformed various computational efforts to win according to the Paper by Google Inc (NASDAQ:GOOGL).
The project was run by Demise Habib’s who described the program as a single general learning system able to combine sensory knowledge and learning in what is known as deep-Q-network in the paper.
Hassabis joined Google Inc (NASDAQ:GOOGL) upon his artificial intelligence company DeepMind being reportedly bought by the search giant for $400 million. He is a well-regarded game designer and a champion gamer. The new findings on the other hand according to researchers underline how far artificial intelligence is from matching with human-type intelligence.
The research team now remains focused on bringing the machine into the 1990’s where it is to learn and navigate far more complicated three-dimensional games like Grand Theft Auto. The machine could be of use on Google Inc (NASDAQ:GOOGL)’s push with the driverless car technology that is expected to be a game-changer in the auto space. Mr. Hassabis affirms that if the machine can drive a car in a racing game then theoretically it can drive a real car.