Part II · Gallery — Symbolic AI

draft

"Good Old-Fashioned AI": intelligence as manipulation of explicit symbols and rules. It dominated from ~1956 to the 1980s. Today it lives in niches and resurfaces in neuro-symbolic hybrids (gallery 06).

🧩 Expert System Symbolic · Knowledge/diagnosis · Rule base + inference engine

  • What it is: captures a human expert's knowledge in IF-THEN rules and

    infers conclusions.

  • Examples: MYCIN (medical diagnosis, 1972), DENDRAL, XCON.
  • Algorithmic basis: forward/backward chaining, production logic.
  • Peak: the 1970s–80s (the commercial "boom" of AI).
  • Capabilities / modes: intellectual/diagnostic; explains its reasoning.
  • → Part IV: chapter planned.

🗺️ Planning and Search Symbolic · Planning/decision · State-space search

  • What it is: finds a sequence of actions leading from an initial state to a

    goal.

  • Examples: STRIPS, PDDL planners, A* in games and routing.
  • Algorithmic basis: heuristic search (A*, IDA*), constraint satisfaction.
  • Peak: the 1970s–present (robotics, logistics, games).
  • Capabilities / modes: intellectual/strategic.
  • → Part IV: chapter planned.

🕸️ Knowledge Graph + Reasoner Symbolic · Structured knowledge · Graph + description logic (OWL/RDF)

  • What it is: represents facts as entities and relations, and infers new

    facts through logic.

  • Examples: Cyc, Google Knowledge Graph, Wikidata + reasoners.
  • Algorithmic basis: description logics, SPARQL/Datalog rules.
  • Peak: the 1980s (Cyc) and the semantic-web revival of the 2010s.
  • Capabilities / modes: intellectual; verifiable and auditable knowledge.
  • → Part IV: chapter planned.

📐 Logic Programming / Solvers Symbolic · Formal reasoning · Resolution / SAT-SMT

  • What it is: solves problems stated as logical formulas or constraints.
  • Examples: Prolog, SAT/SMT solvers (Z3), Answer Set Programming.
  • Algorithmic basis: unification, backtracking, DPLL/CDCL.
  • Peak: the 1980s (Prolog, the Fifth Generation project) and continued use in

    formal verification.

  • Capabilities / modes: intellectual/deductive; guaranteed correctness.
  • → Part IV: chapter planned.

Salient sciences and mathematics in this family: mathematical logic, proof theory, combinatorics/complexity, philosophy (epistemology). It contrasts with the connectionist paradigm by being interpretable and exact, but brittle outside the modeled domain (brittleness).