Explainable Artificial Intelligence and Algorithmic Diplomacy: Enhancing Transparency and Trust in Decision-Making for International Development Institutions

Abstract

This dissertation examines how Explainable Artificial Intelligence (XAI) can be integrated into multilateral decision-making processes through the proposed Explainability Layer for Multilateral Decision Systems (EL-MDS) framework. The study addresses a pressing challenge: while AI promises speed and analytical power for international development institutions, its opaque “black‑box” nature risks undermining trust, legitimacy, and accountability in global governance.
A mixed‑methods proof‑of‑concept design was adopted. The quantitative phase employed simulated data reflecting diverse international contexts (e.g., refugee quota allocations, trade protocols, and debt forecasting) using parameters derived from public datasets. This allowed for rigorous testing of EL-MDS without compromising the confidentiality of high-stake decision-makers. Complementary qualitative interviews with governance experts, NGO officers, and legal advisors were conducted to evaluate the framework’s policy relevance and contextual applicability.
Results demonstrated that EL‑MDS significantly improved stakeholder trust and perceived AI acceptance relative to baseline XAI methods. Simulated participants exposed to EL‑MDS achieved higher mean trust scores (3.2 versus 2.9 on a 5‑point scale) and rated recommendations as more contextually relevant and implementable. Qualitative findings underscored the importance of modular design: the Diplomatic Transparency Engine and Cross‑Cultural Interpretability Layer were particularly valued for aligning outputs with institutional norms.
The research concludes that simulation‑first approaches can effectively de‑risk emerging technologies for use in complex multilateral environments. While findings are indicative rather than definitive, they highlight key priorities for subsequent live pilot studies with partner institutions. Recommendations include co‑design workshops with multilateral organizations, small‑scale pilot deployments, and iterative refinement of modules based on real‑world feedback.
This dissertation contributes a novel framework at the intersection of AI governance and diplomatic practice. It demonstrates how structured explainability mechanisms can strengthen accountability and trust, enabling international development institutions to leverage AI ethically and effectively.
Keywords: Explainable Artificial Intelligence (XAI); Algorithmic Diplomacy; EL‑MDS Framework; Multilateral Decision‑Making; AI Governance; Transparency; International Development Institutions.