AI Reference 1: AlphaGoSpeculative Design
The number of possible scenarios on the chessboard is much larger than the number of atoms in the universe.
Alphago’s search process in the Monte Carlo tree. Source: Nature
For AI, the world is presented in a digital way. AlphaGo includes two deep neural networks: Value Network and Policy Network. The value network makes AlphaGo able to clarify the situation,
the global “strategy”, abandon the inappropriate route; strategic network makes AlphaGo can optimize each step, around the local “tactics” to reduce mistakes. The combination of the two makes AlphaGo do not need too much computing, it can also be out of the exquisite chess game, like humans.
AlphaGo is a sign. It means that people’s exploration of AI has reached a new stage. Creating AlphaGo learning model, will be extended to a variety of areas, such as face recognition, voice recognition and so on. And creating the core of AlphaGo technology, perhaps it helps us in other areas as well.
Alphago is an excellent case of AI, but I feel that there are some hidden dangers that may AI has. If we can be replaced by AI, what is the meaning of our existence? If we already knew that AI is far clever that human beings, do we have elaborative faculty in the future?