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
Rugged State-Space Search: Hill Climbing vs. Simulated Annealing
Solve non-convex multi-modal objective functions via local heuristic exploration.
Click anywhere on the terrain curve to re-seed search agent position.
Annealing Temperature Schedule: T(t) = T₀ × γᵗ
Symbolic Expert System: Propositional Forward-Chaining Logic
Deductive inference on working memory knowledge bases via Modus Ponens.
WORKING MEMORY: TOGGLE OBSERVED FACTS
RULE BASE KNOWLEDGE REPOSITORY
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.