Can Artificial Intelligence facilitate communication within the UN context?

As Artificial Intelligence (AI) technology has become increasingly entrenched in daily practices, it is tempting to imagine its potential to enhance problem-solving. Communication is no exception.

The essence of communication lies in an interactive process where communicators create, interpret, and exchange meanings by using symbols that are influenced by multiple contexts. What happens when the context is the United Nations (UN)? In this specific context, can AI serve as a mediator and/or assistant for interactions involving two or more communicators? Such questions motivate this commentary which summarizes preliminary findings from a research project jointly supported by Tencent and the National University of Singapore under the Technology for Good: Driving Social Impact framework.

Beyond information transmission

Communication may not shine as a standalone priority in the work domain of UN, but it indeed underpins every success of efforts in fulfilling UN mandates as an invisible pillar.

Historically, Claud Shannon’s model defined communication as a process of information transmission from senders to receivers(Shannon, 19481HideShannon, C. E. (1948). A mathematical theory of communication. Bell System Technical Journal, 27(3), 379–423. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x). Similarly, within the UN context, communication serves as an operational tool that ensures timely and effective information flow in different languages to support every stage of internal operations and programme delivery.

Another leading scholar, John Dewey, emphasized that communication was about participation, sharing experience and creating meanings (Dewey, 19292HideDewey, J. (1929). Nature, communication and meaning. In Experience and nature. (pp. 166–207). W W Norton & Co. https://doi.org/10.1037/13377-005). Similar to Dewey’s view, the UN recognizes communication as a strategic function that enhances coordination among UN agencies, builds trust and consensus among diverse stakeholders, and amplifies UN’s efforts worldwide. This is reflected in certain guiding principles that UN staff are required to follow. For example, peace actors must ensure consistency of communication between UN and local authorities to avoid misperceptions that could undermine peacebuilding efforts (UN, 20103HideUN. (2010). UN peacebuilding: an orientation. https://www.un.org/peacebuilding/sites/www.un.org.peacebuilding/files/documents/peacebuilding_orientation.pdf). UN agencies highlight that fostering trust must rely on clear and evidenced-based communication. For instance, the WHO emphasizes the importance of clear communication in building trust with affected communities and in promoting the adoption of life-saving interventions during health emergencies (WHO, 20224HideWHO. (2022). WHO COVID-19 policy brief: Building trust through risk communication and community engagement. https://www.who.int/publications/m/item/who-policy-brief-building-trust-through-risk-communication-and-community-engagement-and-infodemic-management).

In addition, effective communication is essential for the UN to fulfill its role as a convener, knowledge broker and advocate. As early as the 1970s, the influential MacBride Report underscored that effective communication marked by mutual respect could help forge cooperative relationships, reduce tensions, and strengthen mutual understanding while acknowledging differences (UNESCO, 19805HideUNESCO. (1980). Many voices, one world : towards a new more just and more efficient world information and communication order. Kogan Page.). Rooted in such fundamental and long-standing values, it is unsurprising to hear from an anonymous UN officer that UN staff are required to follow the principle of seeking common ground when communicating internally and externally.

Furthermore, communication is also matter of rights and participation within the UN context. To promote information accessibility and inclusion for people with disabilities, 193 countries adopted a UN resolution, advocating easy-to-understand communication practices that empower people with intellectual disabilities to play a leading role in policy development (UN, 20226HideUN. (2022). Promoting and mainstreaming easy-to-understand communication for accessibility for persons with disabilities. In United Nations, Treaty Series (Vol. 1249, Issue 9464). https://docs.un.org/en/A/77/L.37). Kiyo Akasaka, Under-Secretary-General for Communications and Public Information emphasized that communication could mobilize a global response and help unite people and cultures through collective commemorations (United Nations, 20117HideUnited Nations. (2011, April 28). Role of communications not just supportive, but integral to creation of policy, programmes and change, information committee told. https://press.un.org/en/2011/pi1978.doc.htm).

Consequently, UN communication principles are shaped by the way how communication has been conceptualized, and influenced by other factors, including organizational mandates, stakeholders’ unique needs and interests, and global issues. For actors fulfilling UN’s mandates, the ability to communicate complex ideas effectively to diverse audiences is indispensable. In practice, they are required to follow specific principles. For example, information as the backbone of communicative process needs to be multilingual, accurate, impartial, transparent, and accessible. It should be used to foster trust, promote inclusion, strengthen shared understanding, and enhance coordination. In addition, different approaches can be adopted within specific communicative contexts. For instance, gender-responsive approach should be applied to ensure that communication efforts do not perpetuate gender inequalities. In peace operations, a conflict-sensitive approach needs to be taken to avoid exacerbating existing tensions or creating new ones(UN, 2024b8HideUN. (2024b). Strategic communication in peace operations (pp. 1–16). https://resourcehub01.blob.core.windows.net/$web/Policy%20and%20Guidance/corepeacekeepingguidance/Thematic%20Operational%20Activities/Public%20Information%20and%20Strategic%20Communications/Policy%20on%20Strategic%20Communications%20in%20Peace%20Operations%20(2024).pdf). Above all, communication efforts should align with organizational mandates and uphold UN core values, such as equity, respect for diversity, integrity, inclusion, and so on.

Landscape of using AI in UN

Recognizing the transformative potential of AI, UN has increasingly integrated it into knowledge management, programme delivery and internal operations. A UN report identified 716 AI and AI-related projects from over 50 entities across the UN system, with 422 generative AI projects accounting for 34 per-cent of the total (UN, 2024a9HideUN. (2024a). Report on the operational use of AI in the UN System. https://unsceb.org/sites/default/files/2024-11/Report%20on%20the%20Operational%20Use%20of%20AI%20in%20the%20UN%20System_1.pdf). An anonymous UN officer said that AI tools have been increasingly used to assist daily work, from translating, drafting working documents, to analyzing data. In addition, AI has been widely used to assist in retrieving information, analyzing data, and translating multiple languages in the system. The AI integration approach varies across UN entities. Some entities adopted AI based on specific needs of business areas, while others established a leading technology unit to manage AI adoption based on specific organizational needs, governance, and capabilities.

The integration of AI also raises ethical considerations. To better understand the impact of AI, education and training initiatives have emerged in the system. Some UN agencies, such as ITU and UN-OICT, have proactively formed AI-related mechanisms and issued AI guidelines to ensure responsible and ethical use of AI. Normative efforts are also underway, such as the Principles for the Ethical Use of Artificial Intelligence in the United Nations System, and the United Nations System White Paper on AI Governance, to ensure ethical and responsible use of AI in UN.

Facilitating communication elements

Does AI genuinely contribute to improving communication process? In the absence of empirical testing, the results of literature review still offer exploratory insights, indicating the answer is Yes and No.  

Around the same time with Shannon’s model, Lasswell developed a model highlighting five key components – commonly known as 5Ws: who says what (communicator and message), through which channel (medium), to whom (audience), and with what effect (outcome) (Lasswell, 194810HideLasswell, H. D. (1948). The structure and function of communication in society. In L. Bryson (Ed.), Communication of Ideas (pp. 215–228). New York: The Institute for Religious and Social Studies.).

The available literature suggests that AI can facilitate, to varying degrees, communication elements in Lasswell’s model, notably in text-centric and data-driven operations. For example, Large Language Models (LLM) were used to extract and classify unstructured text and images to support decision-making process. Practical examples include WIPO’s use of LLMs to transcribe spoken content into texts in accordance with legal and regulatory requirements; UN Women and UNICC’s use of AI to flag sexist content on social media; and IOM’s use of AI to track immigration sentiment derived from Twitter data during the COVID-19 pandemic (UN, 2024a11HideUN. (2024a). Report on the operational use of AI in the UN System. https://unsceb.org/sites/default/files/2024-11/Report%20on%20the%20Operational%20Use%20of%20AI%20in%20the%20UN%20System_1.pdf). In WIPO’s case, AI fulfills the role of communicators by assisting the message processing. AI also diversifies information sources for communicators by collecting data from social media, as illustrated by UN Women and IOM.

In addition to LLMs, mixed AI models were used to augment communication outcomes. For instance, UNICEF combined AI and Generative AI to create interactive digital textbooks with mixed functionalities, such as sign-language videos, audio descriptions and text-to-speech conversion(UN, 2024a12HideUN. (2024a). Report on the operational use of AI in the UN System. https://unsceb.org/sites/default/files/2024-11/Report%20on%20the%20Operational%20Use%20of%20AI%20in%20the%20UN%20System_1.pdf). AI-generated teaching content enhances the outcome of communication by making content persuasive and engaging for audiences.

Technical limitations

However, challenges such as cultural disparity, linguistic diversity, geopolitical dynamics, and international disputes complicate communication within the UN context. When encountering these challenges, AI may be incapable of providing a solution even if it is willing to.

Taking language and cultural barriers as an example, research indicates that AI’s potential in enhancing language processing and facilitating cross-cultural communication in language learning contexts. By leveraging AI, language learners can gain real-time translation, access to authentic cultural environments, and acquire cultural-specific knowledge (Karakas, 202313HideKarakas, A. (2023). Breaking down barriers with artificial intelligence (AI): Cross-Cultural communication in foreign language education. In Transforming the Language Teaching Experience in the Age of AI (pp. 215–233). IGI Global. ). Hence, researchers argue that AI can improve language learners’ cultural awareness and reduce linguistic and cultural misunderstanding. However, the UN communication context is far more complex than language learning settings. It is not simply about transmission of multilingual messages but also aims to coordinate and build consensus among diverse stakeholders. Therefore, messages must be clear, accurate, consistent, impartial, while remaining highly sensitive to cultural and political nuances.

However, after consulting with technical experts during the project, I found that AI’s performance was far from meeting this requirement. Firstly, technical experts highlighted the limited ability of AI for translation and interpretation. For example, AI translation cannot understand the nuances in languages as quickly as human interpreters. It might be difficult for AI to accurately interpret tones, intended meanings, and emotional nuances in interactions. A technical expert from Tencent said that, currently, AI cannot fully reach the quality of human translation in terms of faithfulness, expressiveness, and elegance. These limitations may result in reduced accuracy. As an anonymous UN officer stressed: “accuracy and impartiality should be prioritized in UN communication process. We cannot allow AI to mislead this process”.

Secondly, AI still struggles to capture the subtleties and cultural nuances of the text as accurately as humans, which does not meet the cultural and political sensitivity requirement of UN communication (Karakas, 202314HideKarakas, A. (2023). Breaking down barriers with artificial intelligence (AI): Cross-Cultural communication in foreign language education. In Transforming the Language Teaching Experience in the Age of AI (pp. 215–233). IGI Global. ; Moneus & Sahari, 202415HideMoneus, A. M., & Sahari, Y. (2024). Artificial intelligence and human translation: A contrastive study based on legal texts. Heliyon, 10(6), e28106. https://doi.org/10.1016/j.heliyon.2024.e28106). Tencent technical experts also noted the inherent cultural disparities in existing large language models resulting from inadequate data training for less commonly used languages and dialects. Moreover, biases in data and algorithms within AI systems can potentially undermine fairness in society, such as in judicial systems (Mehrabi et al., 202216HideMehrabi, N., Morstatter, F., Saxena, N., Lerman, K., & Galstyan, A. (2022). A survey on bias and fairness in machine learning. ACM Computing Surveys, 54(6), 1–35. https://doi.org/10.1145/3457607). Consequently, cultural prejudices in the real world can be reinforced through the application of AI, which contradicts the UN fundamental principles of respecting cultural and language diversity.

Furthermore, with AI increasingly integrated into UN operations, ethical concerns have emerged, especially about how to ensure AI usage aligns with UN core values and fundamental communication principles, such as information authenticity and integrity, privacy, and data security. As commercial AI tools have been adopted within the UN context, assessing inherent risks has become more important than ever.

Conclusion

AI, particularly its language processing ability, has assisted data-driven and text-centric operations in the UN. However, its potential to facilitate cross-cultural communication in non-textual settings remains underdeveloped. Two key technical barriers contribute to this gap. One is AI’s limited ability to comprehend nuances in language and culture. The other barrier is biases in data and algorithms that may undermine social justice and inclusion.

To fully explore AI’s potential in facilitating UN communication, it is essential to ensure that AI applications align with UN core values and organizational mandates. Specifically, messages must be clear, consistent, accurate, while remaining highly sensitive to cultural and political nuances. Ethical concerns about privacy and security must be considered. Communicative efforts must promote trust, inclusion, consensus, and respect for cultural diversity.

Research exploring the value gaps between AI systems and the UN remains very limited. Moving forward, it will be interesting to investigate how AI’s performance can better align with UN’s fundamental communication principles and values, and how this alignment shapes both AI development and UN’s communicative practices.