Skip to content

Knowledge & Reasoning

Representing knowledge explicitly and reasoning over it — the symbolic tradition and its fusion with learning.

Most modern AI learns patterns from mountains of examples, but it can't always tell you why it reached an answer. Knowledge and reasoning takes a different route: you write down facts and rules explicitly, then let the machine follow them step by step — like a board game where every legal move is spelled out, so anyone can check that you played fairly. Store facts such as 'a dog is an animal' and 'Rex is a dog', and the machine can reason that Rex is an animal, showing its work the whole way. This older, logical tradition is precise, inspectable, and hard to fool. Today researchers are fusing it with pattern-learning systems, hoping to get the best of both: the flexibility of learning and the trustworthy, checkable reasoning of logic.

The main ideas

  • Knowledge graphs & ontologies — Structured representations of entities and their relationships (like this very map).
  • Symbolic AI & logic — Rules, logic programming, and search — reasoning you can inspect and verify.
  • Expert systems — Encoding human expertise as rules; a foundational, still-useful approach.
  • Automated planning — Finding action sequences that reach a goal state.
  • Neuro-symbolic AI — Combining neural learning with symbolic reasoning for reliability and interpretability.

Foundations of AI · AI Agents & Autonomy · AI Safety, Alignment & Ethics

Want to make things?

Head to AI School — AI camps where kids build their own games.