I was invited to the BC AI Wildfire Symposium to give insights into the previous and potential uses of Artificial Intelligence and Machine Learning for modeling, prediction and decision making for the improtant and challenging problems raised by Forest Wildfires in British Columbia. This is a Canadian, and indeed, a global problem as climate change is making old models less predictive, conditions are changing more rapidly than ever before and management costs for wildfires are skyrocketing every year. The past two years (2017 and 2018) alone were the worst wildfire seasons in recorded BC history by far, with around 4% of the entire province burning over those two years by some accounts.

Wildfire managers and Forestry researchers have extensive experience managing this system and already have detailed models of fuels, landscapes, ecological impact and fire spread. Other challenges include the logistical planing of moving people, materials and machinary to the right place at the right time. All of this is the context for looking to see if there is more that can be done to improve their data, models and their capacity to manage this complex system.

AI/ML provide an opportunity to provide more powerful tools to do their job better and gain a deeper understanding of the tradeoffs they face every day. I presented an overview of AI/ML and Reinforcement Learning for the diverse group of citizen, operational, business, academic and research stakeholders at the symposium. Several panels then discussed the impacts of and ways to use AI for this domain. It was a highly productive symposium that connected a lot of people that need to communicate more often to solve these big problems.