The very first attempts
The history of artificial intelligence begins in 1956 at the legendary "Dartmouth Seminar" (conference) on artificial intelligence at Dartmouth College.
But one cannot say that this was a completely new direction of technological development. Separate ideas of machine or artificial intelligence have existed before. In fact, the very first attempt at creating an electronic intelligent machine was a machine created during the Second World War to decipher secret military messages. The electromechanical system, then constructed by the British mathematician Alan Turing, can be considered the first prototype of weak artificial intelligence and the first materialized idea of primitive AI.
Initial goals and first methods - the first step
Generation number one. Precise algorithm and technical neural network.
The most important idea of the first generation of Artificial Intelligence was the desire to replace or strengthen the human. It was supposed to be implemented using a computational mechanism with a clear and well-defined set of algorithms. It was not only the period of the formation of the goal but also the time of the technological search for tools to implement the conceived.
Despite the mechanistic nature of the main task, which dominated at that time, it was during the period of work on this concept that the idea of creating an artificial system imitating biological neural groups using an artificial electronic neural network appeared. As a result, the first generation AI appeared at the intersection of clear algorithms and the first computing systems.
Realizing that the main thing is information processing - the second step
Generation number two. Instead of an accurate result and a vertical computing mechanism, information processing becomes the main goal.
In the late 80s and first half of the 90s, a fresh new trend appears, which is quickly becoming the dominant direction for AI development. The main task now is not to solve a specific problem, but to process information (work with a large amount of data). The main mechanism for the implementation of second-generation AI is the so-called flexible algorithms. Now, as the main problem is solved, the parameters of the tool - the algorithm - also change.
We can safely call this generation the most embodied in reality concept. For nearly three decades, most of the developers and programming teams have developed this direction. The use of flexible algorithms has become not just a popular solution, but a real standard in the young AI industry. The most successful and well-known complexes and search engines (for example, Google) have worked on this principle.
But in reality, this was only the beginning of a difficult journey to create AI.
Read the continuation and next steps in the second part of this article.