Symbolic AI represents knowledge in explicit databases and applies handcrafted rules to manipulate it. It dominated AI before machine learning became the leading alternative. section 1.2
A rule-based chess program
An early chess program could use hardcoded rules written by programmers. Because it did not learn from data, it was an AI system without being a machine-learning system. section 1.2
Use the boundary between symbolic and learned approaches as a practical guide:
- Prefer explicit rules for well-defined logical problems, such as chess. section 1.2
- Expect difficulty when rules must capture complex, fuzzy tasks such as image classification, speech recognition, or language translation. section 1.2
- Treat expert systems as a later symbolic-AI revival that became expensive to maintain, difficult to scale, and limited in scope. section 1.11
