It sounds exciting. An artificial intelligence system that learns by itself the intricacies or complex states of video games, and perhaps of anything in its path.
According to Bloomberg, Google’s Deep Mind Technologies, based in London, has created an AI system that can figure out tasks on its own. To start, the company released two animated GIFs showing how the AI system plays Breakout the very first time and how it plays the game a few hundred times later.
The machine-learning technology inside Deep Mind’s AI system was able to complete 49 games played on the retro console Atari 2600, without any specific instructions on how to do so.
A reward system is used after the AI gets a high score or completes certain levels.
The purpose of the AI is not strictly for use in video games. “The ultimate goal here is to build smart, general-purpose machines, but we’re many decades off from doing that,” said Demis Hassabis, co-founder at Deep Mind Technologies and VP of Engineering at Google.
This is the “first time anyone has built a single learning system that can learn directly from experience and manage a wide range of challenging tasks,” he adds.
[Source]: Bloomberg: Google Invents an AI System That Plays Video Games on Its Own.