Artificial Intelligence and Urban Landscape

critical reflections and applications

Abstract

Objective: To analyze the use of analytical and generative artificial intelligence in urban landscape planning and management, evaluating its limits, potentialities, and theoretical and practical implications.

Methodology: Theoretical review based on the critical epistemologies of urbanism, associated with the Design Science Research protocols and practical experiments in two territories of the city of São Paulo, using Roboflow and ChatGPT software.

Originality/Relevance: The study fills a theoretical and practical gap on artificial intelligence applied to urbanism, critically addressing its epistemic, sociotechnical, and ethical dimensions, still little explored.

Results: The analytical dimension showed good results in reading physical and abstract elements. The generative dimension returned partial proposals, with accurate points and relevant errors, requiring high human supervision.

Theoretical/Methodological Contributions: Convergent synthesis of concepts, structuring of a replicable methodology, and advancement in the discussion of the limits of artificial intelligence in urbanism.

Social and Environmental Contributions: Redesigns of urban landscapes according to the parameters of humanized and resilient cities.