Deepmind AI can play multiplayer games, even better than humans


AI has definitely taken over the human brains in almost every field of life. However, we have always restricted ourselves to think that technology is never beyond a ‘few’ human-brain tasks. Like video games.

But here is a twist, deep mind Artificial Intelligence can now play next level multiplayer games without humans and better than humans.

Researchers have proved their excellence with training the AI algorithms to participate and win in the 3D multiplayer game- a modified version of Quake III Arena in Capture the Flag mode. The players in the game are subjected to capture the flag from the enemies, while also protecting their own flag.

As a part of the game, the players are supposed to navigate their way using a map and are to take decisions wisely and stick up to their teammates. The game requires extreme use of co-operational skills and using brainy tricks.

Successfully, AI has aced this game. Players that played with the bot also exclaimed that the bot was very humanly and took all the decisions cleverly.

Earlier this year, 5 bots powered with AI have defeated the most professional e-sports team in the game Dota2. The same case happened when pro players have lost to AI players on StarcraftII. In these cases, a form of reinforcement learning was put in use, and the algorithm learns by trial and error and by interacting with its environment.

These bots were trained while putting them against the clones of themselves in the game. Due to the computer’s speed, the AI could play a game that takes minutes or even hours for humans to play, in a few seconds. This allowed the researchers to coach their AI with 45,000 years of game play inside not even ten months of real-time.

Both the players and the audience have looked at this as a way into the far future.

Tech News Correspondent, Responsible for Writing latest tech new from different domains Internet of things, Artificial Intelligence, Machine Learning, Data Analytics, Big Data, Cloud computing, Data Science, Deep Learning, Online Marketing Strategies and much more. Laasiya can be reached at [email protected]