AlphaGo program is famous for being the first to beat a professional go-to player without a handicap on a standard 19 x 19 board. And this victory is a very significant breakthrough in the field of artificial intelligence, because most experts in this field believed that this event would happen no earlier than 2020-2025 of the year. AlphaGo was taught using the general principles of machine learning and elementary go theory. But more recently, artificial intelligence has learned to do without any help from people and continues to develop completely independently.
The updated AlphaGo version is another important step into the future of machine learning and artificial intelligence. As Nature magazine jokingly notes, such a transition brings us even closer to the “car uprising,” which will acquire superpowers without any help from people. The new version of artificial intelligence was named by the developers of AlphaGo Zero. And it took her only a few days to learn the rules of the game of go from scratch.
At the same time, artificial intelligence came up with its own unique steps, allowing it to easily beat professional go players. On it, even many recognized masters take years. AlphaGo Zero took three days to do it. To test the new AI in action, he was forced to play go with the previous version of the AplhaGo program. As the developers assumed, AlphaGo Zero won with a score of 100: 0. Unlike previous versions, AlphaGo Zero more well limits itself exclusively to human knowledge, and goes beyond them. This is precisely the importance of this moment in the history of the development of artificial intelligence.
The article is based on materials .
Comments
Post a Comment