Often are warning that the development of artificial intelligence could mark the beginning of the end of the human race. However, most of the citizens council approves the intention of large companies to invest in this all the more fruitful field of technology. However, it seems that Google encountered the first signs of resistance to its new program.
It is a program of artificial intelligence called DeepMind, which is 2016, showed great potential. The DeepMind is capable of self-taught, and it is used when training for a performance in the championship game of games “Go!”.
Today, transmitted world media, DeepMind increasingly used a kind of aggressiveness to get to victory in various games.
In recent tests, two DeepMind agent played the game “Gathering” in which wins the player with most collected apples. Programs behaved normally while on the ground were enough apples, but when their numbers began to decline agents have chosen to use a weapon in the game in order to avoid defeat.
“We allow agents to the game play a few thousand times and we let them to independently learn how to behave rationally. Naturally, when the soil a sufficient number of apples agents behave peacefully. But, when the number of apple starts to fall, agents learned that the most cost-effective attack or ‘freeze’ opponents to get to the point “, according to a blog team that studies DeepMind.
With Joel Leib, one of the team members, said that the aggressive behavior of artificial intelligence shows some aspects of human behavior that occur as a product of environment and learning.
“Less aggressive behavior occurs when the resources are sufficiently accessible. The motivation of greed appears as a reflection of the temptation that rivals can not afford to keep apples for themselves,” Leibo said.