This is neither the first match between humans and AI (Artificial Intelligence), nor the first time that AI defeats a human being. In 1997, IBM Computer, Deep Blue makes chess history by winning Garry Kasparove, the chess grandmaster. Another computer developed by IBM, called Watson, has steadily defeated former winners Brad Rutter and Ken Jennings and became a winner in the quiz show Jeopardy !.
Nonetheless, the match between AlphaGo and Lee Sedol was the most sensational event of all time. The reason for this is that the abstract strategy board game named Go is considered one of the most intellectual mind games that creates a challenge of extraordinary depth and complexity. What did DeepMind do as an important AI breakthrough to make AlphaGo the first computer against a human Go player?
AI needs training like a human being. In the past, humans have set the patterns and activated the AI. However, in recent years AI has created the patterns themselves based on data. That’s what we called Machine br> Learning. According to Tom M. Mitchell’s Machine Learning book, machine learning is the study of computer algorithms that automatically improve through experience.
Cloud System and AI
Unlike Deep Blue and Watson, AlphaGo is based on a cloud computing system. With the parallel connection of many CPUs and GPUs becomes a large cloud computing system, which acts as a human brain to the AI. It’s easy to expand resources and improve performance than a single supercomputer. All experiences, the most important element of machine learning, are also stored in cloud computing. Jeff Dean, a Google Senior Fellow in the System Infrastructure Group, said Google is using a cloud computing system for Google services like Google Search, Google Photo and Gmail.