Artificial Intelligence: Origins and Definition

Artificial Intelligence Origins

• As a research area, Artificial Intelligence is 62 years old.
• The term was defined in a summer workshop in 1956.
• The workshop spanned 6-8 weeks.
• It was held at Dartmouth College in New Hampshire, U.S.
• The workshop occurred shortly after the first computer's creation.
• Early computer science work focused on low-level tasks.
• Original ambitions for computer use were delayed.
• The workshop aimed to revisit these original goals after a decade of preliminary work.

Defining AI

• The goal is to find how to make machines that:
• Use language
• Form abstractions and concepts
• Solve problems reserved for humans
• Improve themselves
• A selected group of scientists can significantly advance on these problems by working together.
• John McCarthy made this statement at Dartmouth College in 1956.

Founding Figures in AI (1956)

• Claude Shannon: founder of Information and Communication Theory.
• D.M. Mackay: British researcher in Information Theory and Brain organization.
• Julian Bigelow: chief engineer for the von Neumann computer at Princeton in 1946.
• Nathaniel Rochester: author of the first assembler for the first commercial computer.
• Oliver Selfridge: named "the father of Machine Perception".
• Ray Solomonoff: inventor of Algorithmic Probability.
• John Holland: inventor of Genetic Algorithms.
• Marvin Minsky: key MIT researcher in the early development of AI.
• Allen Newell: champion for symbolic AI and inventor of central AI techniques.
• Herbert Simon: pioneer in Decision-making theory and a Nobel Prize winner.
• John McCarthy: inventor of the LISP programming language.

Early AI Work (1950-1960)

• Alan Turing wrote "Computing Machinery and Intelligence" in 1950.
• In 1955, Allen Newell, Herbert A. Simon and Cliff Shaw created the Logic Theorist was the first AI program.
• It could prove theorems in Whitehead and Russel's Principia Mathematica.
• It introduced list processing, means ends analysis and heuristic search.
• John McCarthy created Lisp in 1958.
• Lisp is based on Lambda Calculus and uses list processing.
• Lisp is the second-oldest high-level programming language still in use.
• Only Fortran is older.
• Oliver Selfridge created the Pandemonium architecture in 1959.
• Pandemonium was one of the first computational models in pattern recognition for images.
• In 1959, Simon, Newell, and Shaw created the General Problem Solver (GPS).
• GPS is a computer program to work as a universal problem solver machine.
• Any problem expressed as well-formed formulas (WFFs) or Horn clauses can be solved by GPS.

Early Machine Learning Work (1950-1960)

• Arthur Samuel coined "Machine Learning" in 1959.
• He pioneered machine learning with programs that played checkers at IBM's Laboratory in 1956.
• McCulloch and Pitts introduced Neural Networks as a model of computation in 1943.
• Marvin Minsky and Dean Edmonds built SNARC in 1956.
• SNARC was the first Neural Network machine able to learn.
• Frank Rosenblatt invented the Perceptron in 1957 at the Cornell Aeronautical Laboratory.
• John Holland introduced Genetic Algorithms in 1960.
  • Genetic Algorithms are based on Darwin's evolution theory.
• Ray Solomonoff published work on Non-semantic machine learning in 1956.
• His first contribution was called an Inductive Inference Machine.
• He published work on Algorithmic probability in 1960.