- Author:admin
- Time:2016-12-17 19:56
Recently, two students of the United States use deepness study technology, made an AI system, can be not only in classical game Doom oneself play game, still can be opposite in much person the human rival that it beats in battle.
Those who develop this technology is Devendra Chaplot and Guillaume Lample, 2 people come from card to be able to bear or endure Jimeilong university, the nickname of the robot that they train in game is called Arnold. The Gu Ge before this uses AI system in the 2D Atari game that restore ancient ways, and these two students went to AI body application in 3D fire game.
Resemble human player same, AI body also is to repeat ceaselessly play this game, arrive how to kill mate to study all the time, computer pilot part includes already among this, also include part of human player pilot.
Actually, also a few other groups did similar thing with Doom, but what these two students differ is, should send a paper to discuss its to implement a method in detail accordingly, and the paper is in a person of the same trade to discuss level at present.
They mention in the paper: “ normally, deepness study method just uses visual input to train. We used a kind of method that enhances training model, will dig game more photograph to pass information, for instance the enemy is mixed the characteristic information of article. Our framework also undertook modular, can let different model, in the different level of game, independent ground undertakes training. ”
This technology makes a person impressive undoubtedly, criticism also followings sb's heels and come however, somebody thinks this technology also trains a homicide machine effectively, may turn prospective robot into the foundation of killer.
“AI is becoming truer and truer. Although so far, AI body still is only move in world resembling element, but the problem ” that it caused the AI in real world to develop really, say of futurology home Dom Galgeon, we do not think “ to say AI the thing of exceeding danger, but still need to think seriously make clear and sound AI policy, include its research, development and application. ”