Summary

Recent advances in artificial intelligence (AI)—largely based on machine learning— offer possibilities for addressing climate-related security risks. They are particularly useful for addressing risks related to climate hazards and risks related to climate vulnerabilities and exposure. This SIPRI and Mistra Geopolitics policy report explores the possibilities and challenges AI can bring as well as examples of how AI can be used for climate security.

Key messages

• AI provides three possible opportunities for climate security: understanding and predicting the impact of climate hazards, managing vulnerabilities and exposure to climate change, and detecting climate change-related grievances and tensions.
• To illustrate these opportunities, researchers discuss examples such as conflict early-warning systems (CEWS), which have incorporated environmental and climate change-related factors as inputs for forecasting. The Water, Peace and Security (WPS) partnership, for instance, has its primary focus on addressing water-related conflict risk by generating conflict forecasts.
• Even though AI can assist with data and enable efficient use of remote sensing systems for policy, especially satellite imagery and social media, the use of AI for climate security presents technical and ethical chal­lenges. Among the challenges AI might bring is ensuring that the system is trained on representative and reliable data. This is critical to ensuring the usefulness of the system and to minimizing the risk of bias.
• There is a need to consider associated ethical concerns, such as a policy intervention that discriminate against certain social groups—particularly those that do not engage with social media for economic, social or political reasons.

Conclusion

Based on these key findings, policymakers interested in further exploring the potential of AI for climate security should support critical research on AI and climate security. They should also support access to digital infrastructure and digital literacy, and the develop­ment of an open access AI tool for the collection of climate security-related data in conflict-affected and fragile countries.

Researchers who wish to make use of AI to better understand the nexus between climate change and security should conduct empirical research and explore methodological questions. They should also consider ethical and political risks associated with use of AI methods to monitor climate change-related tensions and political grievance. In all this, they should maintain close links with affected communities.

Publisher