AI Challenge
AI Challenges aim to bring together multidisciplinary teams of policymakers, diplomats, foreign policy practitioners, data scientists to develop cutting-edge solutions that can potentially assist and augment the work of foreign policy professionals.
NegotiateCOP – Supporting Climate Negotiations
International negotiations, particularly within the framework of the United Nations, are complex and challenging, especially in times of geopolitical tension and a multipolar world order. This is particularly true for the negotiations on the Paris Agreement and the annual climate conferences, which have been held since the first UN climate meetings in the 1980s and the first COP in 1995. The aim of these conferences is to prevent or mitigate the impacts of the climate crisis by limiting global temperature rise, reducing greenhouse gas emissions, and implementing measures for climate change adaptation.
Prototype: AI Avatar – A trainer for diplomatic negotiation skills
This innovative tool is designed to support diplomats or attachés in training or in training in developing and practicing their negotiation skills. Users have the option to select from a range of predefined opponents representing diverse roles and areas of expertise, such as a diplomat specializing in climate policy, a Big Tech entrepreneur, or an EU representative focused on tariffs. Each avatar articulates its perspective in alignment with its role and strategic interests. Users can respond by recording their own statements, to which the avatar replies according to its position. Beyond predefined scenarios/roles, this prototype also offers the option to create customized cases: users can upload documents that the system processes into a new position or avatar for dialogue and debate. By engaging with these structured yet dynamic interactions, trainees can practice their responses /strengthen their ability to respond to complex real-world scenarios.
Leveraging LLMS in enhancing multilateral negotiations
November 2023 - January 2024
This project, conducted in partnership with Omdena, leveraged cutting-edge Large Language Models (LLMs) and advanced techniques to address challenges faced by foreign policy experts in navigating complex negotiations. Specifically, we developed a Proof of Concept AI assistant to help policy officers summarize a large number of policy documents and make queries from these documents. Focusing on the UN's Global Digital Compact (GDC) negotiations as a test case, the project successfully illustrates the benefits of integrating AI tools into the workflows of policy officers. We are delivering a Proof of Concept (PoC) system suitable for beta testing, which is also a solid foundation for a widely deployable production system should further development work be pursued.