A computer program that is capable of teaching itself? Sounds like a dream. But it is actually a reality. AlphaZero is a program that learns while from its mistakes and keeps learning new things.
Until now, we have seen computers act out on the instructions that are given to them. These instructions are given to them in their coding. But what if we make a program that is coded to learn new things?
Is it possible to have a computer program that is capable of teaching itself? It sounds like a dream. But it is a reality. AlphaZero is a program that learns while from its mistakes and keeps learning new things.
AplhaZero consists of a single system that can master a game from scratch. AplhaZero grasped the games of Chess, shogi (Japanese Chess), and Go within very little time. After learning these games, AlphaZero was also able to beat the World Champion Garry Kasparov.
After his defeat to the AI, Kasparov said, “I can’t disguise my satisfaction that it plays with a very dynamic style, much like my own!”
This ability to learn each game afresh, unconstrained by the norms of human play, results in a distinctive, unorthodox, yet creative and dynamic playing style. Chess Grandmaster Matthew Sadler and Women’s International Master Natasha Regan analysed many games of Chess played by AlphaZero.
The experts analysed its playing style and said that ” it is unlike any traditional chess engine.” The same experts further added, “It’s like discovering the secret notebooks of some great player from the past.”
A program that learns on its own uses the method of observing and remembering its own mistakes. It also creates solutions to the mistakes and registers the right moves. So if you win against the robot using one strategy, then the next time, the robot will not let you win using the same.
You too can make a similar program like AlphaZero, which learns on its own. Similarly, a robot that is capable of learning on its own can be the way ahead in Artificial Intelligence. This will help in making better versions of humanoid robots too.