Symbolic AI is the approach to artificial intelligence that represents knowledge with explicit symbols and logical rules, reasoning by manipulating those symbols rather than learning patterns from data.
Symbolic AI is also called Good Old-Fashioned AI, or GOFAI, and dominated the field from the 1950s through the 1980s. It encodes the world as discrete symbols — entities, properties, relations — and operates on them using formal logic, search algorithms, and hand-authored rules. Classic instantiations include expert systems, theorem provers, semantic networks, knowledge graphs, and logic programming languages such as Prolog. The defining commitment is that intelligence emerges from explicit, inspectable representations rather than statistical pattern recognition.
The approach contrasts sharply with the connectionist or sub-symbolic tradition that underpins modern 📝Machine Learning (ML) and deep learning, where knowledge lives in the weights of neural networks and resists direct inspection. Symbolic systems offer transparency, formal verifiability, and strong compositional reasoning, but historically struggle with perception, ambiguity, and the brittleness that comes from manually encoded knowledge — the difficulties that drove the AI winters of the 1970s and 1980s. The current revival takes the form of neuro-symbolic AI, which attempts to combine the learned representations of neural networks with the structured reasoning of symbolic systems.
Symbolic AI remains the historical and conceptual counterweight inside the broader 📝Artificial Intelligence (AI) field — the side of the discipline that bet on logic, structure, and explicit knowledge rather than scale and statistics.