Symbolic AI, also known as “Good Old-Fashioned Artificial Intelligence” (GOFAI), refers to the approach in artificial intelligence research that emphasizes the use of symbols and rules to solve problems. Unlike neural networks and other machine learning approaches that rely on statistical patterns, symbolic AI focuses on logic and symbolic reasoning.
In symbolic AI, knowledge is represented through symbols, such as words or images, and rules that dictate how those symbols can be manipulated. These rules can be expressed in formal languages like logic, enabling the system to perform reasoning tasks by following explicit procedures.
The symbolic approach has been particularly useful in domains where clear, rule-based reasoning is essential, such as in expert systems, natural language processing, and certain types of problem-solving.
While symbolic AI was the dominant approach in the early years of AI research, it has since been complemented and, in some areas, surpassed by statistical and data-driven methods. The reason for this shift is partly due to the difficulty in encoding complex real-world knowledge and common sense into a symbolic form, along with the rise of powerful machine learning algorithms that can learn from large data sets.
However, symbolic AI still has its place in hybrid systems that combine rule-based reasoning with statistical learning, creating more robust and explainable AI models.