Google found two A sit together to play the game they will work or fight

Google found two A sit together to play the game they will work or fight

Category : yibpgtal

believes that since Alfa beat the dog, many people want to simply two artificial intelligence directly to the final accounts.

Google of the Deepmind team also did a similar thing, but also a little bit of a slightly different, this is due to the test objective is not to find which artificial intelligence playing chess is more tired harm, but, artificial intelligence in particular some rules of environment, what is the cooperation is also



after all, the future of artificial intelligence in this society, it is possible to deal with a lot of things, small to command traffic, big to the country’s economic operation, may be involved in artificial intelligence. However, the history of human development tells us that sometimes cooperation may be the best for everyone.

to investigate this issue, DeepMind did an experiment, said several AI robot into a series of "social dilemma", in these circumstances, individuals can benefit from the betrayal of others, but if everyone is so selfish, there is no winner. Does this remind you of the "prisoner’s dilemma" in the "game theory" put forward by Nobel laureate?

Deepmind to do the experiment is…… Let the AI play two simple video games together.

the first video game called picking apples, two players in the middle of a tube on the collection of apples, who picked up more apples who win. At the same time, they have a skill, by blocking the action of the other side of the laser, so that opponents temporarily disappear in the game, the former and the opportunity to collect more apples.

second a game called "wolves", hunt two game player needs to be full of difficulties and obstacles in an environment of a prey. However, it is not only grasping the game that the players can score, but when the prey was arrested, all players close to the prey can score.

experimental results are slightly interesting, but it is not surprising that these artificial intelligence will be based on different scenarios and rules to adjust their behavior, cooperation and sometimes hostile.

, for example, in the "pick it up" in the game, when the number of apple is very abundant, these artificial intelligence seems to be too lazy to do hinder others, but to pick up the apples in a big. However, as the total number of apples gradually decreases, the number of laser skills they use increases. Liberal arts students may begin to show in the YY what kind of people in extreme cases of the most authentic human nature, what Marx’s original theory of communism was broken……

but this is a robot, the next thing is even more interesting, when the introduction of a more powerful artificial intelligence operations, but he is inclined to no matter how many apples left, will hinder opponents. It seems that the more intelligent AI is showing, in this game, when to be aggressive.

but does that mean that AI thinks that blocking others is the most

Leave a Reply

Recent Comments