Individual Submission Summary
Share...

Direct link:

Projecting Conflict Under Climate Change: An Artificial Intelligence Application

Thu, August 29, 10:00 to 11:30am, Hilton, Jefferson East

Abstract

In recent years, the research community has witnessed a growing interest in forecasting the outbreak, occurrence or severity of violence. Yet, forecasting efforts have been mostly focused on the socio-economic and institutional predictors of conflict, and only limited research has explored the role that climatic and environmental conditions may play in mitigating or fostering violence in the future. Although climate change is not likely to have a direct influence on the risk of conflict, changing climatic conditions have been proven to interact with well-established drivers of violence, including socio-economic development, natural resources, food security and human health. The present study represents the first attempt to project the probability of conflict in the long-term future, under changing climatic and environmental conditions. The main focus is on the effects that climate change and agricultural production may have on conflict risk. To increase the predictive power of the analysis, we apply and compare the results from two Artificial Intelligence derived tools: Random Forest and Artificial Neural Networks. In the context of Artificial Intelligence, Machine Learning techniques could overcome the limitations of conventional forecasting tools based on statistical analyses and take advantage of an increasing availability of data. The inclusion of climatic and environmental predictors in the analysis, coupled with this innovative methodology, could enhance our ability to forecast violence and identify some key conflict hotspots in future scenarios.

Authors