Brief history of AI

Original article was published by Pavel Ilin on Artificial Intelligence on Medium

Dartmouth workshop

Main approach of the AI becomes a formal logic approach. The center of AI research became the US because the rest of the world was in ruins or lacked technological and educational foundation.

Term AI was proposed by John McCarty in 1956 during the Dartmouth workshop. The workshop was not a focused project, rather exploratory discussion which covered many subjects. During the workshop new methods were created, such as symbolic methods, earlier expert systems and so on.

The same time, computer hardware keeps advancing and more mathematicians and philosophers begin to get access to it. Computers are still mainly used for military purposes, but gradually, during the nights or weekends non military researchers were able to play with it.

Age of optimism

The Golden Age of AI took place between 1956 and 1974. It was an age of great optimism. Corporations and the government invest a lot of resources into the field. Everyone thought that we need to push a bit more and we will have real AI which can solve the most complex tasks humanity faces.

Sci-fi literature picked up and heated up an optimistic vibe. Such authors as Azek Asimov, Arthur Clark, Stanislav Lem were creating worlds with sentient AIs.

During this period AI field developed in several main directions.

Math logic and fuzzy logic

With math logic methods researchers were able to create a program which can prove some theorems. Alan Roberdson proposed an algorithm of checking deductive thinking.

On the other hand Fuzzy logic was developed and it’s widely used in neural networks nowadays. Fuzzy logic is a form of many-valued logic in which the truth values of variables may be any real number between 0 and 1 both inclusive. It is employed to handle the concept of partial truth, where the truth value may range between completely true and completely false. By contrast, in Boolean logic, the truth values of variables may only be the integer values 0 or 1.

Intelligence as a search

We can create trees of all possible combinations and teach machines to find optimal solutions. For example you can map all possible moves in the tic tac toe game and teach machines to try all possible options.

This approach was able to show the general public some interesting results. Machines learn how to play chess, checkers and so on.

Natural language Processing

In parallel there was another field in development, NLP. Researchers were full of hope that we can formalize human language and translate it to the code. And if there was a success with syntax, AI lacked the cultural context based on which languages a developed. For example AI failed spectacularly with the task of translation from one language to another.

Interesting event of that time was development of the first ever chat bot ELIZA which was created in 1966. It was a simulation therapist. ELIZA examined the text for keywords, applied values to said keywords, and transformed the input into an output; the script that ELIZA ran determined the keywords, set the values of keywords, and set the rules of transformation for the output.

Neural networks

Frank Rosenblatt lays the foundation of the neural networks fields. He created the first neural computer (Mark1) which was able to recognize handwritten text.

Robotics

The same time, first industrial robots became a reality. However they are not connected with the AI field directly, but it’s a field which converges with AI. You need to control robots and make them do what you need and they have their own “brains”.

First AI winter