Part II · Gallery — Evolutionary AI

Intelligence through simulated natural selection: populations of solutions that mutate, recombine, and compete across generations. Inspired by evolutionary biology and collective behavior.

🧬 Genetic / Evolutionary Algorithm Evolutionary · Optimizationsearch · Population + selectionmutation/crossover

• What it is: evolves a population of candidate solutions guided by a fitness

  function.

• Examples: NASA antenna design, schedule optimization, evolutionary

  generative art.

• Algorithmic basis: selection, crossover, mutation, elitism; evolution

  strategies (CMA-ES).

• Peak: the 1970s (Holland) – present (black-box optimization).
• Capabilities / modes: creative optimization; explores without gradients.
• → Part IV: chapter planned.

🧠🧬 Neuroevolution Evolutionary + connectionist · Controlgames · Evolution of network topologyweights

• What it is: uses evolution to discover architectures and weights of neural

  networks.

• Examples: NEAT, HyperNEAT, evolution of game agents.
• Algorithmic basis: evolution of network graphs, novelty search.
• Peak: the 2000s–2010s; resurfacing in neural architecture search.
• Capabilities / modes: architecture discovery without backprop.
• → Part IV: chapter planned.

🐝 Swarm Intelligence Evolutionary · Distributed optimization · Simple agents + local rules

• What it is: intelligent behavior emerges from many simple agents

  interacting locally.

• Examples: PSO (particle swarm), ant colony (ACO), swarm robotics.
• Algorithmic basis: local rules, stigmergy, population-based optimization.
• Peak: the 1990s–2010s (routing, logistics).
• Capabilities / modes: collective/distributed; robust to failures.
• → Part IV: chapter planned.

Salient sciences and mathematics: evolutionary biology, complex-systems theory, optimization (gradient-free), probability, game theory. Strength: explores non-differentiable and multimodal spaces; weakness: computational cost and the absence of guarantees.