How can the UN ensure the impartiality of its AI platforms?

As the UN celebrates its 79th birthday on October 24, it grapples with a mix of familiar and novel challenges. From financial constraints to political blockage in the Security Council, these issues have always been a test for the organisation. Now, the advent of artificial intelligence (AI) poses a new and potentially profound threat, jeopardising the UN’s core principle of impartiality and potentially reshaping its operations.

This risk for the UN is confirmed by the recent research of Belgian and Spanish scientists who analysed ideological biases of 17 popular LLMs in both English and Chinese. Their comprehensive study of a moral and ideological judgement of 4,229 political figures worldwide shows clear biases reflected in developers’ views of specific LLMs platforms.

The forthcoming AI transformation is not new for the organisation. The UN previously faced policy and security dilemmas, from a shift to cloud computing in the 2010s to online meetings during the COVID-19 pandemic. However, unlike previous technologies, AI is not just a passive tool; it can shape policy outcomes. This makes the transparency and impartiality of the UN more critical than ever.

The challenge: How to ensure the UN’s impartiality

Impartiality is what lends legitimacy to the UN, ensuring that decisions are based on facts and diverse perspectives. As the UN explores AI for reporting, drafting, and analysis, it faces a crucial question: how can it ensure the impartiality of its AI platform?

Proprietary AI systems risk embedding biases that reflect the values of their creators, potentially skewing how the UN Secretariat approaches global issues. This is a fundamental threat to the fairness and objectivity that the UN must uphold.

The solution: Open-source AI

To counter this, the UN must adopt an open-source AI approach—one that is transparent, inclusive, and collaboratively developed by countries, companies, and citizens. By building an open-source AI platform, the UN can ensure that the technology aligns with its core values of impartiality, inclusivity, and transparency, drawing on global expertise through rigorous evaluation.

AI@UN systems must be explainable and traceable, ensuring that data and algorithms are accessible and understandable. The ‘weights’ assigned to data and information should reflect core UN values, avoiding biases that could favour one country over another.

AI@UN: a paradigm for AI as a public good

This moment presents both a challenge and an opportunity. By committing to an open, transparent, and inclusive AI framework, the UN can transform a potential risk into a landmark opportunity for global governance. This isn’t just about adopting new technology—it’s about reaffirming the UN’s capacity to evolve, lead, and remain relevant in a changing world.

By building AI@UN, the organisation can set a global standard for AI governance—prioritising human rights, inclusivity, and the public good over other interests. AI@UN represents a chance to address these challenges and lead by example, developing AI that truly serves the public good.

New opportunities to get back to the core mission

AI holds immense promise for the UN. By freeing diplomats and UN officials from repetitive paperwork, the organisation can focus more on its core mission: negotiating peace and promoting the global public good. AI can, in particular, empower small and developing countries by levelling the diplomatic playing field, allowing them to engage more effectively in multilateral processes.

Next steps

The UN and its member states should embrace the development of a transparent, open-source AI@UN platform. By doing so, the UN can ‘walk the talk’ of Agenda 2030 and the Pact for the Future. It can transform AI’s risks into opportunities for greater inclusivity, impartiality, and effectiveness, reinforcing the UN’s role as a beacon of hope and leadership on the global stage.

Happy birthday to the UN and to everyone who believes in the organisation’s critical role for the future of humanity!