Skip to main content

History of AI

How old is thinking machine idea?​

can be dated back to ancient times, when Greek myths and legends portrayed mechanical beings endowed with intelligence.

Who/when do we credit with exploring mathematics of how brain operates?​

René Descartes

1st AI philosopher​

Ada Lovelace

When was AI term adopted by academics for area?​

1956: Dartmouth Conference

GPS​

General Problem Solver: describe (model) problem in logic and it will solve it using logical inferences

→ You cannot make essentially represent anything in the world via a logic? The problem is just too big, and the real world problems are too fuzzy.

What type of AI existed before AI-Winter I/what about that type of AI didn’t succeed as expected?​

GPS

AI Winter I​

No one want to fund AI projects anymore (AI Winter I)

What type of AI succeed after AI Winter I before AI Winter II?​

Knowledge-based approaches:
• 1969—79: Early development of knowledge-based systems
• 1980—88: Expert systems (path finding) industry booms

→ Only solved too specific problems

What types of AI have succeeded post AI Winter II?​

Statistical approaches, machine learning, AI Spring, Deep learning, neutral network

What are some pivotal events post AI Winter II in AI?​

  • 1997 Deep Blue beets Gary Kasparov
  • 2006: UoT develops Deep Learning
  • 2011: IBM's Watson won Jeopardy
  • 2016: Go software based on Deep Learning beats world's champions

Connectionist​

  • inspired by the Human Brain
  • systems of interconnected units (or "neurons") that work together to process information.

Deep Learning​

  • networks consist of multiple layers of nodes (hence "deep" learning) that process information in a hierarchical manner.
  • type of connectionist

Compare connectionist AI to symbolic AI.​

  1. Connectionist
  • Based on neural networks, inspired by the human brain.
  • Learns from data through training focusing on pattern recognition, classification, and prediction.
  1. Symbolic
  • Based on logic and explicit symbol manipulation
  • Rule-based systems that follow predefined logical rules

What are some future/currently developing AI areas from class slides?​

  • Future: Deep learning, machine learning, digital assistants, service architectures, internet of things, self-automation