Deep Blue
IBM's chess computer that beat world champion Garry Kasparov in 1997. Included here as a contrast: it won by fast search and hand-tuned rules, not by learning.
Also IBM Deep Blue · Kasparov · game tree search
IBM, 1997
A calculator for chess. It looked at an enormous number of possible moves very quickly and scored positions with rules written by experts. It did not learn by itself.
Where it fits
+All relationships as a list
How it works
Deep Blue is in this atlas mainly as a contrast. It was a triumph of search, not of machine learning.
In May 1997, an upgraded Deep Blue beat world champion Garry Kasparov 3½ to 2½ in a six-game match, the first time a reigning world champion lost a match to a computer under standard tournament time controls. Its strength came from:
- Brute speed: custom chess chips evaluating on the order of 200 million positions per second.
- Expert knowledge: an evaluation function and opening book built and tuned by people.
Nothing in it learned from experience the way a neural network does. Note that its search was alpha-beta search, an older relative of the Monte Carlo tree search that later programs used. That approach hit a wall in Go, where positions are too hard to score with hand-written rules; AlphaGo succeeded there in 2016 by learning its intuition and combining it with search.
Everyday use cases
- Case 01
How you've probably met it
As the famous 1997 headline "computer beats world chess champion", a landmark moment in public perceptions of AI.
Strengths, weaknesses and when to use it
Strengths
- Showed raw search plus expert knowledge could beat the best human at chess.
Weaknesses
- Did not learn and did not transfer; its approach failed for games like Go, where learned intuition (AlphaGo) was needed.
When to use itAs a historical contrast between hand-built search and learned systems.
Compare with
Go deeper
Sources
Facts checked as of 2026-10-12