Frequently Asked Questions: AI in the UN System
A comprehensive FAQ guide about how Artificial Intelligence is being implemented and managed across the United Nations system, including challenges, frameworks, and future recommendations.
Published 2025-01-10 · By Shahzad Asghar
Frequently Asked Questions: AI in the UN System
How is AI being used within the UN System?
AI is being adopted across various UN agencies to enhance their operations and contribute to sustainable development. Examples include using predictive analytics for humanitarian response, automated language translation for multilingual communications, and data analysis to inform decision-making in areas like agriculture, urban planning, and disaster response. AI is also being used to create chatbots for information access and support various back-office tasks. The breadth of applications highlights the transformative potential of AI in supporting the UN's mandate.
What are the main challenges in adopting AI across the UN System?
The decentralised nature of the UN system poses challenges. Individual agencies often develop AI solutions independently, leading to duplicated efforts, missed opportunities for collaboration, and inefficient resource allocation. The complex, interdisciplinary challenges faced by the UN also necessitate collaborative, inter-agency AI development, which can be difficult to foster. Furthermore, ensuring the ethical, responsible, and secure use of AI is a key concern given the sensitive nature of the UN's work and the diverse contexts in which it operates.
What mechanisms are in place to promote knowledge sharing and collaboration on AI within the UN System?
Several networks and working groups facilitate knowledge sharing and collaboration. These include the High-Level Committee on Management (HLCM) Digital and Technology Network (DTN), the HLCP Inter-Agency Working Group on AI, the UN Innovation Network and its Generative AI Practice Group, and the ITU AI for Good Platform. These groups organise meetings, share guidance, and create tools and frameworks to support AI adoption. Additionally, the UN is actively cataloguing AI projects, use cases, and open-source software to make resources accessible and encourage collaboration.
What is the PRISM framework, and how is it being used?
The PRISM framework, developed by the DTN GenAI Community of Practice, is a tool designed to help UN organisations evaluate and prioritise AI use cases. It uses a structured approach based on several key criteria to assess potential AI applications. These criteria include alignment with organizational needs, potential for risk reduction, non-financial value, and technical feasibility. This framework helps ensure that AI initiatives are well-aligned with organizational goals and resources and that the adoption of AI is a strategic decision.
How is the UN system addressing the risks associated with commercial AI offerings?
WIPO has developed a risk assessment framework specifically for evaluating commercial AI offerings. This framework helps UN organisations identify and mitigate potential risks during the planning and implementation phases of AI projects, covering areas such as data security, bias, and regulatory compliance. The framework aims to support agile risk assessments, recognizing that AI offerings are rapidly evolving and require proactive measures to ensure responsible implementation.
What role do open-source practices play in the UN's approach to AI?
Open-source practices are increasingly important in the UN's approach to AI. The UN recognises the benefits of open-source software in promoting transparency, collaboration, and cost-effectiveness and is actively encouraging the development and use of open-source tools and resources. Several UN initiatives, like the DTN Open-Source Software Community of Practice, are working to establish a UN open-source code-hosting platform, standardise licensing, and build capacity in open-source practices. The intent is to ensure the UN is contributing to, and not merely consuming, these vital resources.
What is Retrieval Augmented Generation (RAG), and how is it being used in the UN system?
RAG is a technique used in Generative AI applications, particularly chatbots. It works by augmenting user queries with relevant information from a knowledge base, allowing the AI to provide responses grounded in specific contextual information. This technology is being adopted by many UN organizations to enhance their information management, knowledge sharing, and decision-making processes. Examples include ITCILO's AnswerMate, which provides internal governance information.
What are the key recommendations for the future of AI in the UN System?
Key recommendations include: centralising and standardising the cataloguing of AI activities and use cases, promoting knowledge sharing through AI communities of practice, using frameworks like PRISM for prioritisation, supporting the development of open-source AI resources and projects, investing in UN-led AI development, and actively engaging in wider AI platforms and forums. Furthermore, strengthening collective bargaining agreements with AI technology vendors, creating internal policies for AI governance, and investing in in-house AI expertise are also important steps. The overall goal is to ensure the responsible, effective, and collaborative adoption of AI across the UN System.