History of Artificial Intelligence

(Master of Information Science) Lecture on Pattern Recognition

• History of Artificial intelligence
  • In the early days, research on Chess (tasks that resemble intelligence) was the main focus.
    • Tasks like image recognition were thought to be easy because humans can do them easily.
  • In the 1980s, there was a surge of excitement around the idea that “if we can define the rules, we can solve any task!”
    • For example, detecting faces using rules defined by humans.
  • However, it turned out that defining those rules was difficult.
    • The goal was to replicate human perception, but it was unclear how humans perceive things in the first place.
  • After that, there was a period of stagnation, and the approach of feeding large amounts of data became popular.
    • Various datasets were created.
    • (blu3mo) It can be said that this was a step up to a more advanced (meta) way of thinking.
      • TOK-like, thinking about how to learn rather than what to learn.
    • (blu3mo) Well, even before the meta approach, the essence lies in the fact that artificial intelligence is “learning.”#cybernetics
  • Just when things were not going well, Deep Learning emerged (the technology itself already existed, perhaps originating from Japan?).