In recent years, researchers from different companies are actively developing this promising technology, like artificial intelligence. Google is one of the undisputed leaders in AI race and all the forces working to improve his artificial intelligence DeepMind. In one recent study the company’s engineers have trained an artificial intelligence to overcome various obstacles in a virtual environment, why born a kind of virtual traceur.
To achieve outstanding results to researchers, Google has allowed the so-called method of “reinforcement learning”, when artificial intelligence is inside a certain environment, which he does not yet have any details, but gets the opportunity to take some action. By trial and error, artificial intelligence exploring their environment, gradually understanding and accepting the rules by which there is a virtual world in which he found himself.
We invite you to see the video of the experiment that demonstrates the achievements of DeepMind still a new art of parkour. On some of the shots impossible to watch without a smile, but it’s still a very good result for artificial intelligence Google.
#video | Google taught artificial intelligence.
Sergey Grey