AI@UN: Navigating the tightrope between innovation and impartiality
The UN faces various risks, from funding shortages to geopolitical deadlock in the Security Council. AI introduces new risks, challenging the UN’s ideal of impartiality and potentially altering its operational approach.
As the UN adopts AI for reporting, drafting, and other core activities, its impartiality will be tested. Most current AI platforms are proprietary and opaque, falling short of the UN’s high standard for neutrality. Such systems could shape the UN Secretariat’s thinking and influence Member States’ decisions.
An open-source AI platform, developed with contributions from countries, companies, and citizens, offers a way forward.
The following outlines conceptual and practical steps for could be named AI@UN.
Why is impartiality important for the UN?
The principle of impartiality is the linchpin of the UN’s legitimacy and credibility, ensuring that policy advice remains objective, grounded in evidence, and sensitive to diverse perspectives.
Impartiality is the linchpin of the UN’s legitimacy and credibility, ensuring that policy proposals of the UN Secretariat are objective, evidence-based, and sensitive to diverse perspectives.
The concept of bureaucratic impartiality traces back to the British civil service, which inspired the establishment of international administration and the role of international civil servants. The Northcote-Trevelyan Report of 1854 laid the foundation for a professional civil service, emphasizing impartiality.
This principle is even more crucial at the UN, where impartiality among governments can determine the organization’s success or failure.
AI and the evolution of technology at the UN
AI represents a new chapter in the UN’s history of leveraging technology. From typewriters during the League of Nations in the 1930s to telegraphs for communication, the UN has consistently adopted new technologies.
In October 1963, the UN used cutting-edge satellite technology to connect New York with the ITU in Geneva, where Secretary-General U Thant addressed the first satellite communication conference.
In the 2000s, digitalisation at the UN accelerated through two significant shifts: the transition to cloud-based data storage in the 2010s and the move to online meetings in the 2020s.
Shift to the cloud
In the early 2010s, the UN moved its datasets and archives from internal servers to the cloud, raising questions about technical and legal protections.
Technically, once data moved off UN premises, its protection fell to cloud providers like Amazon, Microsoft, and Google.
Legally, diplomatic immunity became a key concern—how to ensure that tech companies upheld the immunity of UN data.
Luxembourg saw an opportunity in this ambiguity, offering “digital embassies” with data stored locally on private servers under international diplomatic law. Estonia and the ICRC soon established such embassies in Luxembourg. As tech companies seek more data to train AI models, protecting UN data remains crucial.
Protecting UN data will receive new relevance as tech companies rush to get more data to train their AI foundational models.
Shift to online meetings
The COVID-19 pandemic prompted a second shift—from physical to online meetings. While UN buildings ensure security and immunity for in-person gatherings, using private platforms like Zoom and Microsoft Teams raised concerns about the legal status of online diplomatic venues.
Some countries faced restrictions in using conference apps due to sanctions, threatening the UN’s impartiality and inclusion. The UN secretariat had to work on solutions to ensure access for all Member States.
As diplomacy returned to physical meetings, the urgency to define the legal status of online gatherings lessened, but the issue remains unresolved. A proposed solution is a UN open-source online meeting platform, developed as a “public good” by companies, countries, and citizens.
Why is AI different from previous technological innovations at the UN?
Unlike past technologies, from typewriters to Zoom meetings, AI can actively shape arguments and influence outcomes in UN policy debates and negotiations. Off-the-shelf AI systems carry inherent biases from the data and algorithms on which they are developed, raising questions about their impartiality. Ensuring impartiality requires full transparency and explainability throughout the AI cycle, from the foundational data to the assignment of weights within the system.
‘We, the peoples,’ the first line of the UN Charter, should guide the development of AI at the UN. Contributions from countries, companies, and communities worldwide can enhance AI’s potential to support the UN’s missions of maintaining global peace, advancing development, and protecting human rights. An inclusive approach to AI development is essential to uphold the UN’s principle of impartiality.
An inclusive approach to AI development is key for upholding the principle of impartiality, a pillar of the UN.
The AI@UN has two main goals:
- Support policy discussions on the sustainable AI transformation of the UN ecosystem.
- Inspire member states and other actors to contribute AI models and agents.
Guiding Principles for AI@UN
1. Open Source: Follow open-source community principles and ensure transparency at all stages of the AI lifecycle, including curating data and knowledge for AI systems, selecting parameters and assigning weights to develop foundational models, vector databases, knowledge graphs, and other segments of AI systems.
2. Modular: Develop a family of AI agents in a modular way by following shared standards for improbability. Start with AI agents for core UN activities such as reporting, event organisation, and negotiations.
3. Public Good: Use AI to codify UN knowledge as a public good for countries, communities, and citizens. Inspire similar efforts to codify other knowledge sources, including ancient texts and oral traditions. ‘Walk the talk ‘ of public good narratives.
4. Inclusive: Encourage contributions of AI models and applications from member states, companies, and academia, similar to cultural artifact donations to the UN offices.
5. Multilingual: Represent diverse linguistic and cultural traditions, focusing on capturing and preserving wisdom from oral traditions not found in written texts.
6. Diverse: Ensure contributions from a wide range of perspectives. Maintain transparent traceability of sources behind AI-generated outputs.
7. Accessible: Adhere to high accessibility standards, especially for people with disabilities. Ensure simple, low-bandwidth solutions to make the system affordable and inclusive.
8. Interoperable: Overcome knowledge and data silos within the UN using ontologies, taxonomies, data curation, and shared technical standards.standards.
9. Professionale: Follow high ethical and industry standards for planning, coding, and deploying AI solutions. Use peer review to ensure reliability, with a focus on human impact.
10. Explainable: Trace AI-generated outputs to their sources, ensuring transparency and impartiality at the highest possible technical level.
11. Protecting data and knowledge: Safeguard AI systems’ data, knowledge, and inputs by protecting privacy and intellectual property rights. The first step is to ensure the explainability and traceability of AI answers to their data and knowledge sources.
12. Secure: Guarantee high security and reliability. AI communities will be encouraged to contribute to red-teaming and security tests of the AI@UN system.
13. Sustainable: Support the SDGs by prioritizing them in model development. Ensure AI systems are sustainable through resource sharing, proper documentation, and environmentally conscious practices.
14. Capacity development: By developing an AI system, the UN should develop its own and wider AI capacities. Efforts should be holistic, involving the UN Secretariat, Member States, and other communities, and comprehensive, covering basic to advanced AI skills.
15. Future-Proofing: Prepare for emerging technologies like augmented reality and quantum computing using the experience and expertise gained from AI@UN.
Opportunities in crisis
AI transformation will inevitably create tensions as it impacts deeper aspects of how the UN functions. Likely resistance, rooted in human fears and attachment to the status quo, should be addressed openly and reframed to highlight the opportunities AI brings to individual officials, Member States, and the UN as a whole.
The AI transition presents, among others, the following opportunities:
First, AI can empower small and developing countries to participate in UN activities in more informed and impactful ways. By compensating for their limited diplomatic missions and resources, AI can help these nations keep pace with larger systems, reducing the current imbalance in AI capabilities.
Second, AI can support the UN Secretariat in reallocating time and resources. By automating traditional paperwork, such as report preparation, AI can free up valuable capacity for fieldwork in Member States where the UN’s assistance is most needed.
Call for action
By developing an open-source AI@UN platform the UN will be able to evolve, take the lead, and remain relevant in a rapidly changing global landscape.
By leveraging the transformative power of AI, the UN can turn a looming challenge into a watershed moment, ensuring the organisation’s relevance and leadership in charting the course of human progress for all.