AI-Driven Discovery Reveals Critical Thresholds and Persistent Inequities in Urban Sustainability

Cities worldwide aim to simultaneously achieve environmental protection, social equity, and economic vitality. These seemingly simple goals, however, require evaluating sustainability trade-offs across hundreds of instances, projects and places – a cognitive limitation traditionally mistaken for physical impossibility. In this paper, we present a multi-objective AI framework that analyzes hundreds of census tracts and variable configurations across three competing objectives to achieve maximum performance (in all three) simultaneously (+ 90%). The solution sets of such a large and complex array have been computationally invisible to conventional analysis. Using the city of Chicago as our test case (801 census blocks, 59 configurations) we found 22 uniquely optimal solutions that cluster along century-old green space corridors established by Burnham's original 1909 Plan. Interestingly, these solutions are found in the wealthier north and northwest parts of city with none in the highly disinvested south and west. This pattern demonstrates how historic infrastructure decisions can create path dependencies that may be difficult to overcome. Our AI framework reveals critical thresholds that appear to enable solution success, including critical open space access and population densities. The work provides an evidence-based approach for climate-responsive urban design and planning decisions. The framework supports an emerging urban cognitive ecosystem approach that uses data and AI to create intelligent, adaptive ecosystems that can learn and proactively deliver services, transforming planning from pattern recognition to revealing systemic and historic inequities.