On Oct. 18, 2017, Google unveiled its most advanced and powerful AlphaGo series product – AlphaGo Zero. Two years ago, DeepMind, the innovative AI company affiliated with Google, invented the first generation of AlphaGo, which was the first computer that could beat the human champion on this ancient Chinese game. Today, the second-generation product, AlphaGo Zero, only used 72 hours to learn everything by itself, beating the last generation AlphaGo, and becoming the smartest computer.
This brand-new AI product learns everything by repeating and playing with itself. Systematically analyzing the chess board it played with its “brother.” We can see a lot of similar routines that have been used in the past classic games, however, how the AlphaGo used them is impressive.
Although AI has become so intelligent now, it was merely a fantasy 60 years ago.
- AI as a thought – Shannon
AI has a much longer history than we know. In 1949, Claud Shannon, the father of modern System Theory, published an article on how a computer might be made to play chess, and people started to discuss whether a computer could match human intelligence.
2. AI as a code – Samuel
Since the year that Shannon put forward the idea of the computer playing chess, Samuel spent seven years developing the game of checkers, and it became the starting of AI exploration and machine learning. It was also this innovation that set the foundation of AI – in the future research and experiments, how well the computer can play chess is the standard of how intelligent the computer is.
3. AI as a winner – Chinook
In 1989, Jonathan Schaeffer and his team at the University of Alberta innovated a Checkers program, called “Chinook,” the first computer that completely dominates humans at a complicated game.
After Chinook, AI technology stepped into a brand-new era, with the Internet and the development of semiconductor, amazing creations were made. We will talk about the next period of AI history in the next blog. See you next week!