The History of AI

(in less than 60 seconds)

AI 0.0 = Human generated Rules (Before 2000)

Humans codified their knowledge and intuitions into Rules, and programmed machines to implement these rules.

"Dear Machine, If your opponent played this move, then play that move."

AI 1.0 = Learning from Data (After 2000)

Progress in computer hardware, and the ability to collect large amounts of data, led to Statistical Machine Learning on Data.

"Dear Machine, Here's a history documenting millions of chess games. Try and learn how to play chess.

AI 2.0 = Learning from Data, generated from Simulation (After 2010)

Further progress in hardware, led to fancy simulation software. Simulation led to vast amounts of data, which in turn could be used to do more elaborate statistical machine learning (also running on fancier hardware).

"Dear Machine, You can learn how to play chess by playing millions of games your self"

AI 3.0? (After 2020)

Simulating complex behaviour, including human behaviour?

“Dear Machine, This stuff is too complex to get my head around. Can you run zillions of simulations and see how things will turn out, and then tell me what to do?