Machine Perception and Cognitive Robotics Laboratory

AgentLife: Artificial Life Meets Agent LLMs

AgentLife builds artificial life ecosystems on a cellular automata substrate where each cell is an LLM agent. Agents undergo mutation, selection, and crossover, and develop their own communication protocols and languages. Think Conway's Game of Life, but every cell can think.

Model Card

Cellular automata grid where each cell is an LLM agent. Evolutionary operators: mutation (prompt perturbation), selection (fitness-based survival), crossover (context merging). Agents develop emergent communication protocols.
No external training data. Agents evolve behaviors through interaction on the CA substrate.
Agents develop specialized communication protocols. Population dynamics show classic evolutionary patterns — speciation, niche formation, arms races.
LLM inference cost per cell per timestep limits grid size. Evolutionary timescales require many generations.

Team

William Edward Hahn, Ph.D.

Founder and Co-Principal Investigator (Co-PI)
Associate Professor

[email protected]

Machine Perception and Cognitive Robotics Laboratory