Robots created for logic games with humans do not impress the average layman too much. However, one should not rush to conclusions. The fact is that it is these programs that today demonstrate the most impressive results in self-learning, self-improvement and imitation of human thinking.
For example, the artificial intelligence "AlphaGo", which will be discussed in this news, became famous in May this year for being the first to beat the world professional in Go - a popular board game with deepest strategic content. And he did it without any giveaway from a person. This victory was a very significant breakthrough in the field of AI, as scientists well versed in Go predicted that it would not occur until 2025.
Previously, for training "AlphaGo" used the general principles of machine learning and elementary theories of the game in Go, but recently the robot refused any help from its creators and began to develop completely independently.
Someone will be surprised: what is so outstanding about this, and how can this program be called advanced artificial intelligence? They say that the algorithm goes through many options for placing stones on the board and finds optimal solutions for winning. Delov… Nevertheless, in fact, everything is much more complicated and interesting.
AlphaGo works on the basis of neural networks, that is, it imitates, by and large, human mental activity. In fact, the program does not enumerate any options, since their number for the game of Go is so great that even the most powerful modern supercomputer cannot cope with this task in millions of years. Artificial intelligence uses logic and a kind of semblance of intuition, it also bluffs and defames purely humanly. Playing with people, he constantly studies human behavior, borrowing clever moves from living players, remembering unsuccessful ones, possibly absorbing even human emotions.
The program has already learned to come up with its own unique methods of playing, allowing it to easily beat the world champions. Renowned masters have to hone their strategy for many years in order to develop such steps. It takes AlphaGo a matter of days. Unlike other programs designed for logic games, AlphaGo uses not only human knowledge, but also goes far beyond it. Some seasoned Go players admit that the program plays "at the level of God," and you can compete with it not in order to win, but to learn the skill from it. So, dear skeptics, who do not believe that artificial intelligence will soon catch up with human intelligence, and then surpass it!
