Artificial Intelligence and Peacebuilding: Opportunities and Challenges

The report examines the dual-use nature of AI in peacebuilding, highlighting its potential to enhance conflict forecasting, real-time mapping, and inclusive dialogue while warning against risks such as algorithmic bias, surveillance, and the erosion of human facilitation skills. It emphasizes that AI's effectiveness depends on a healthy information environment and a rights-based approach to design and governance.

• AI tools are being used for conflict prediction and early warning, with systems like ACLED's Conflict Alert System (CAST) reportedly achieving accuracy rates of up to 95% for forecasts up to six months in advance.
• AI-enhanced mapping and satellite imagery are significantly accelerating humanitarian response; for example, the DISHA system has shown potential to reduce damage assessment time by a factor of six.
• AI-powered platforms like Remesh and Polis are scaling citizen engagement in conflict zones, such as in Libya, Yemen, and Syria, by facilitating large-scale dialogues and identifying areas of consensus.
• There is a critical risk of 'dual-use' where AI tools intended for peacebuilding are repurposed for military targeting and surveillance, as seen with the Israeli military's use of the Gospel and Lavender systems in Gaza.
• The report identifies a significant 'implementation gap' where an abundance of AI-generated data does not automatically translate into policy action due to complex political and economic realities.
• AI's role in mediation is currently most effective in a support capacity—such as automating administration or scenario planning—rather than as a substitute for human mediators in high-stakes conflicts.
• The sustainability of AI peacebuilding initiatives is threatened by precarious funding and low user engagement, exemplified by the DEEP platform where only 20% of registered users show active annual participation.
• A healthy information environment—characterized by digital literacy, diversity of voices, and accountability—is a prerequisite for AI to support peacebuilding rather than exacerbate conflict.
• Current Large Language Models (LLMs) continue to suffer from hallucinations, with error rates between 0.8% and 5% even in ideal conditions, which poses dangers in critical humanitarian operations.

APA

Zelizer, C., Ogenga, F., Schirch, L., Tauchnitz, E., Howard, P. N., & Valenzuela, S. (2025). Artificial Intelligence and Peacebuilding: Opportunities and Challenges. International Panel on the Information Environment. https://www.ipie.info/research/tp2025-3

Chicago

Zelizer, Craig, Fredrick Ogenga, Lisa Schirch, Evelyne Tauchnitz, Philip N. Howard, and Sebastián Valenzuela. Artificial Intelligence and Peacebuilding: Opportunities and Challenges. International Panel on the Information Environment, 2025. https://www.ipie.info/research/tp2025-3.

Wikipedia

{{cite report |last1=Zelizer |first1=Craig |last2=Ogenga |first2=Fredrick |last3=Schirch |first3=Lisa |last4=Tauchnitz |first4=Evelyne |last5=Howard |first5=Philip N. |last6=Valenzuela |first6=Sebastián |title=Artificial Intelligence and Peacebuilding: Opportunities and Challenges |publisher=International Panel on the Information Environment |date=2025 |url=https://www.ipie.info/research/tp2025-3 |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{zelizer2025artificial, author = {Zelizer, Craig and Ogenga, Fredrick and Schirch, Lisa and Tauchnitz, Evelyne and Howard, Philip N. and Valenzuela, Sebastián}, title = {{Artificial Intelligence and Peacebuilding: Opportunities and Challenges}}, institution = {International Panel on the Information Environment}, year = {2025}, url = {https://www.ipie.info/research/tp2025-3}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }