Session Overview Climate change a pressing problem Uncertainty around future climate change as currently visible impacts happening faster than previously anticipated Rapid advancement in AI with more available data, more sophisticated algorithms, and more computation power Significant climate data available to train AI models with Needs for AI models for climate change: Models capable of capturing finer details Ability to adapt with changes with the future Good accuracy for making the model usable for the real-world Decisions understandable to various stakeholder without AI background
Few Potential Impacts of Climate Change Warmer temperatures More severe weather More wildfires Sea level rise Coastal erosion Changes in ocean temperature / composition Loss of biodiversity Food insecurity More diseases Displacement Poverty Drought Infrastructure Damage
Collaborative Ecosystems Collaborative Ecosystems of different expertise are needed to make AI useful for real-world applications Few selected roles in the collaborating teams
Key Discussion Points AI Applications: What kind of AI models are being worked on to solve the climate problem(s)? How close are these to make an impact in the real-world? Data and Applicability: What climate data is abundantly available for training, and what data is there a lack of? How do we validate the models work in the real-world? Collaborative Ecosystems: What expertise do the different collaborators bring to these programs? How were these collaborative ecosystems formed? Transparency and Explainability: How are these models explainable and transparent to stakeholders without AI background? Long-term Support: Are the models capable of adapting as the impacts of climate change vary overtime? Which experts will be involved in the long-term maintenance of the models? Legal and Regulatory Aspects: Who owns the AI models, validates, and updates them overtime? Should the models be available to all for public service, or would they be private?