Classical AI Sandbox

Classical AI Sandbox: Non-ML Artificial Intelligence Lab

Symbolic & Search Lab

Adversarial Game Tree: Minimax & Alpha-Beta Pruning

Deterministic game-theoretic backward induction without neural networks or statistical training data.

Legend: Green border = Evaluated node | Red dashed = Pruned sub-tree | Green stroke = Optimal minimax path Click any leaf circle to increment terminal payoff

Classical Symbolic AI vs. Modern Machine Learning

Why AI is a rigorous, expansive discipline that exists completely independent of neural networks.

Classical & Symbolic AI (GOFAI)

Operates on explicit logic rules, state graphs, and provable mathematical bounds. Zero training sets needed.

  • Guarantees: Exact minimax optimality, completeness, verifiable truth.
  • Explainability: 100% white-box proof traces and state auditability.
  • Core Toolkit: A* Search, Alpha-Beta, Constraint Satisfaction (CSP), First-Order Logic, Planning.
  • Sample Domain: Deep Blue chess, NASA Mars Rover planners, medical diagnostics.

Statistical Machine Learning (ML)

Approximates functions by optimizing weight matrices over large empirical datasets.

  • Guarantees: Empirical risk minimization, statistical generalization.
  • Explainability: Black-box latent weights; prone to hallucination.
  • Core Toolkit: Gradient descent, Transformer attention, Backpropagation, CNNs/LLMs.
  • Sample Domain: Computer vision, generative natural language, speech recognition.