GOFAI, or Good Old-Fashioned AI, is a term often used to describe the classical approach to artificial intelligence that emerged in the mid-20th century. This approach emphasizes the use of symbolic representations of problems and logical inference as the primary mechanisms for simulating intelligent behavior. GOFAI systems are built on a foundation of predefined rules, where the world is represented through symbols (such as words or numbers), and the AI manipulates these symbols according to formal logic to draw conclusions or make decisions. These systems excel in areas where problems can be clearly defined and rules can be explicitly programmed, such as in expert systems or theorem proving. However, GOFAI has limitations in handling more complex, real-world environments where tasks require learning, perception, or dealing with uncertainty, which are better addressed by modern AI techniques like machine learning and neural networks.
GOFAI's roots trace back to the 1950s and 1960s, during the early days of AI research, when symbolic AI was the dominant paradigm. The term "GOFAI" itself was popularized in the late 1980s by philosopher John Haugeland, who contrasted it with newer AI approaches like connectionism. GOFAI saw its peak in popularity during the 1960s through the 1980s but began to wane as machine learning and neural networks rose to prominence in the 1990s.
Key figures in the development of GOFAI include John McCarthy, who coined the term "artificial intelligence," Allen Newell, and Herbert A. Simon, who developed the General Problem Solver, one of the first AI programs. John Haugeland, a philosopher, was instrumental in defining and popularizing the term GOFAI, especially in critiquing its limitations compared to newer AI methods.