Summary

Deep Blue was a computer developed by IBM to play chess at grandmaster level. It made history in 1997 by defeating the then reigning world champion Garry Kasparov. Deep Blue used advanced computing power and algorithms to analyze possible moves and choose the best course of action.

ELI5

Imagine you like to play the game of chess, and you have a very clever friend who helps you. This friend is so good at playing chess that they can even beat a world champion! Deep Blue was just like this clever friend, but instead of a person, it was a computer built by people at a company called IBM.

In-depth explanation

Deep Blue was a chess-playing computer developed by IBM. It is considered a landmark in Artificial Intelligence due to its ability to beat a world champion chess player, Garry Kasparov, in 1997. Deep Blue is not an AI in the modern sense, as it did not learn to play chess from data but followed an explicitly programmed strategy.

The system was developed as a massively parallel, RS/6000 SP Thin P2SC-based system with 30 nodes, each containing a 120 MHz P2SC microprocessor for a total of 480 cores. Each node employed multiple specialized chips to speed up the move calculation process, calculating approximately 200 million positions per second.

Deep Blue’s strategy was mainly based on the minimax algorithm, a decision-making algorithm for minimizing the worst-case scenario in a game. It evaluated billions of chess board configurations to decide the optimal move. Deep Blue was not only about brute force calculation capabilities; its programmers also coded a vast number of chess scenarios and positions into the system, effectively teaching it strategic gameplay.

Resulting from a long series of advancements and prior systems like Deep Thought and its predecessors at Carnegie Mellon University, Deep Blue represents an important milestone in the evolution of AI. It signified a shift in how researchers could approach AI and problem-solving systems, influencing the development of AI systems designed for other games (e.g., Texas hold ’em, Jeopardy!, and Go) and practical applications beyond gaming.

Minimax Algorithm, Artificial Intelligence, Machine Learning (ML),, IBM Watson, Distributed Computing, Game Theory, AlphaGo