��ݮ��Ƶ

Angel Hsu
Machine learning, generative AI, including Large Language Models (LLMs) hold immensepotential in addressing critical data and information gaps within city climate policy. Thesetechnologies enable cities to estimate greenhouse gas emission sources, identify environmentalhotspots, evaluate policy performance trends, and comprehend the diverse impacts of climatechange. This presentation will showcase several case studies that illustrate how AI-drivenapproaches can innovate data collection, analysis, and policy formulation in the context of urbanclimate management, including evaluating net-zero climate policy and strategy with LLMs andtopic modeling, distributional climate and environmental impacts within urban areas, theintegration of satellite remote sensing for participatory heat-stress mapping, among otherexamples. It will also discuss some of the challenges and pitfalls of applying AI to urban climatepolicy and management, such as the "black box" AI problem and biases of underlying trainingdata.

Dawn Parker & Rodrigo Costa
This presentation focuses on methods to harness city climate data to develop policy solutions.
While emerging AI methods are exciting, often for policy analysis, we need to design scenarios
that ask how specific policy levers might impact social, economic, and climate metrics in our
cities. To conduct such analysis, we need models of how shifts in policy levers change the
decisions of key actors, and how these decisions interact to drive outcomes of interest. From a
science viewpoint, these are often referred to as “process-based models.” Such models can
complement, and build on, pattern-based AI and other statistical models. In this presentation, we
will offer a brief introduction to several process-based simulation models that can be applied to
climate challenges in cities: systems dynamics, cellular automaton, and agent-based models.