The 1950s and 1960s were the period when artificial intelligence was primarily concerned with the development of computer programs that could perform tasks that were considered to require a high degree of intelligence, e.g. games such as chess; theorem solving etc. A key development during this period was the idea of heuristics, an important precursor to the advent of expert systems. Heuristics can be defined as guidelines for choosing among alternative actions. They can be used as shortcuts to direct the search for a solution along more promising lines, even if an optimal solution is not guaranteed. Another key development was the creation of LISP, a symbolic programming language.

Artificial intelligence began to address broader aspects of intelligence during the late 1960s and early 1970s. Research was oriented toward modeling cognition, interpreting natural language, story understanding and ways to represent and reason about diverse kinds of knowledge. This was also the time period in which artificial intelligence was applied toward solving practical, real-world problems. Examples of these applications were:

Dendral , a program to assist in determining the molecular structure of chemical compounds

Macsyma, a program which could solve mathematical problems symbolically.

Strips, a program which could construct plans to achieve some goal, as an ordered series of steps.

These programs could operate at near expert levels, but still relied to some extent on search techniques rather than on a knowledge base.

The Modern Period

---

The expert systems explosion of the late 1970s and early 1980s was caused by the realization that computer programs could perform useful tasks at expert levels of performance, if they were endowed with large amounts of specialized knowledge, and were constrained to narrow but real domains. Research in this period turned toward trying to clone human experts by capturing their experiential knowledge. Some of the successful expert systems of this period were:

Mycin, a computer program designed as a decision aid for doctors, which when given data describing a patient's symptoms could diagnose infectious blood diseases and prescribe therapies appropriate to the disease diagnosed.

R1, an expert system used by Digital Equipment Corp., which when given a set a specifications of the computer system requirement of a customer would select the appropriate computer components and peripherals, check for inconsistencies, design the layout of the entire system and print out a detailed order.

Prospector, a program to detect commercially viable ore deposits based on geological data.

These successes led to the idea of an expert system that had the basic structure in which rules could be entered, and the matching capability to make inferences based on the rules. The simplicity of this concept led to the rapid commercialization of expert systems. Thus, it was with the start of the early 1980s, that knowledge-based systems were applied to a wide variety of areas. [Early] [Cognitive] [Modern]

---

---