Modeling heat transfer in an urban settlement with 3D cellular automata and artificial intelligenceLeskovar, Robert (Avtor) cellular automataurban heat transferhuman-machine interactionAIassisted codingurban microclimateThis chapter presents a three-dimensional (3D) cellular automata (CA) model simulating urban heat transfer in an artificial 50 × 50 × 20 m city block at moderate latitude (� = 46.24◦ N). Each cell is assigned one of five material classes — outdoor air, asphalt, park/grass, structural concrete, and interior air — and advances temperature states through eight physics layers: conductive diffusion, solar heating, nocturnal radiative cooling, free-atmosphere relaxation, asphalt plume convection, surface Newton cooling, interior ventilation, and boundary conditions. A distinguishing aspect of this work is its development workflow: rather than conventional coding, human-specified intent and domain constraints in natural language guided Claude AI (Anthropic) [27] in generating, debugging, and iteratively refining the Python implementation. The human supplied physical intuition, validated outputs, and corrected physically implausible results through a joint debugging process. Preliminary simulations over a 30-day May period demonstrate realistic diurnal temperature cycles, street-surface temperatures 4– 6 ◦C above park surfaces at solar noon, and an asphalt plume decaying exponentially up to 10 m above ground. These results confirm the CA paradigm’s suitability for urban heat studies [5, 2] and illustrate the potential of large language model (LLM)-assisted scientific prototyping.20262026-07-13 12:29:37Članek v reviji98847UDK: 004.8OceCobissID: 281943555COBISS_ID: 283876355DOI: 10.18690/um.fov.5.2026.8sl