By Javier Surasky
.png)
Why AI Matters in the Succession to António Guterres
Artificial intelligence is reshaping every area of the United Nations’ work, as well as the functioning of the UN itself, but those seeking to succeed António Guterres at the head of the Organization do not seem to attach the same importance to it or offer equally developed responses.
In this blog, I analyze what the candidates for Secretary-General say, and what they leave unsaid, in their vision statements and during the interactive dialogues before the General Assembly, in order to determine what place AI occupies in their proposals.
Based on these sources, I reconstruct the general orientation of each candidacy regarding the international governance of AI, its effects on development, peace and security, employment, digital divides, and the internal functioning of the United Nations, in an analysis in which I distinguish explicit references from general formulations and documentary silences; understanding that the absence of a developed position is, in itself, a position insofar as it allows us to approach the degree of priority and elaboration that each candidate assigns to the issue.
The study is deliberately limited to the types of primary documents produced within the formal framework of the selection process: the vision statements, submitted together with the candidacies, and the interventions in the interactive dialogues convened by the General Assembly, in such a way that comparison on a common basis across all candidacies becomes possible, even if the price we pay is leaving aside possible references to AI in previous speeches, interviews, and actions carried out in other positions.
Rebeca Grynspan: AI as Structural Transformation
and societies, militaries, information systems, the nature of work itself.
AI occupies an explicit and cross-cutting place in her vision statement, where it is presented as a transformation affecting the economy, society, work, information systems, and military capabilities, and she proposes incorporating AI tools and data analysis into the UN’s operational capacities, particularly in early warning and conflict prevention.
This integration was also present in the interactive dialogue, where Grynspan highlighted the capacity of algorithmic decisions to produce cross-border effects: “The algorithm created in one country reshapes lives in every other.”
Grynspan offers the most structural conception of AI among the candidates considered here, where this technology is incorporated as part of a transformation of international relations that requires common governance.
Following several years of her leadership at UNCTAD, Grynspan’s position expresses the recognition of an asymmetric technological interdependence in which the impacts of AI are distributed internationally, while the capacities to design it, control it, and benefit from it remain concentrated. Yet she does not do this directly, but rather by emphasizing the need to avoid the expansion of digital divides and by pointing to the risk that innovation may reinforce preexisting inequalities.
In her approach, the role of the United Nations would consist fundamentally of expanding the dissemination of technological capacities in order to democratize them and ensure that their benefits reach governments, communities, and young people who could otherwise be excluded from the digital world.
As for the institutional adoption of AI by the UN, she proposes that the Organization use AI to analyze risks, strengthen prevention, and improve its operations, as well as lead a process of building global governance, thereby assigning it a dual role as facilitator of regulation and as user of AI systems.
However, this position presents some unavoidable tensions: the rapid incorporation of algorithmic tools for conflict prevention may improve the capacity to process information, but it may also introduce problems of bias, explainability, data quality, and accountability. Neither the statement nor the Costa Rican candidate’s participation in the dialogue specifies how those systems should be audited, who would be accountable for their errors, or how the capacity for challenge by affected States and communities would be preserved.
Her approach may be described as broad and politically significant, but more centered on diagnosis than on the architecture of AI governance: it identifies the transnational and distributive nature of the problem, but does not define precise mechanisms to regulate technology companies, democratize access to infrastructure, or oversee the use of AI within the Organization itself.
Rafael Mariano Grossi: An Instrumental and Marginal View
The Argentine does not develop a position on artificial intelligence in his vision: centered around the idea of the effectiveness of the United Nations, the reduction of duplication, crisis prevention, active diplomacy, and the need to demonstrate results, it neglects this indispensable front for achieving what he proposes as a basic objective, an absence that shows that technological governance does not autonomously form part of his priorities.
By contrast, Grossi did make an explicit assessment when responding during the interactive dialogue, when, in answer to a question from Trinidad and Tobago about his position in relation to harnessing AI to reduce development gaps within and between Member States, he pointed out that the IAEA already interacts with AI and with companies in the sector, mentioned the Global Digital Compact, and expressed a general view of AI as a technology whose impact depends on how it is used.
Grossi’s position seems to be framed within an instrumental kind of technological optimism, in which AI does not appear as a structure that redistributes power, generates new dependencies, or modifies the conditions of political action, but is considered mainly as a neutral tool: for Grossi, the key problem is not the control of infrastructures or inequality in the capacity to decide which criteria are included in an algorithm, which places AI within an administrative and pragmatic logic.
The thesis of technological neutrality raises difficulties, because it fails to see that AI systems do not emerge in a political vacuum and shifts critical decisions about models and algorithms into the realm of “uses,” when a significant part of their effects is, in fact, embedded in design and in the material conditions of access. As a result, Grossi does not develop positions on corporate concentration, technological dependency, infrastructure gaps, autonomous military systems, data protection, or responsibilities, nor does he manage to define what role the UN should play vis-à-vis the private actors that concentrate frontier capacities.
Macky Sall: AI, Development, and the Digital Divide
In his vision document, Sall places AI among the great contemporary technological transformations and presents it as an opportunity to promote development, although he does not provide elements for an architecture oriented toward its governance that go beyond general statements.
During the dialogue, Sall developed his position on AI with greater clarity, noting that “Artificial intelligence should enable us to narrow [the digital divide] further” and tying it closely to elements that drive development.
From a view favorable to technology, although with benefits conditioned on the establishment of governance guided by prudence, he presents AI as a tool capable of reducing gaps and improving the functioning of the UN, differentiating himself from those who propose a scarcely critical acceptance of innovation and affirming that technological progress requires political direction, education, and international cooperation.
However, he leaves only general elements when it comes to defining redistributive policies that would prevent the expansion of AI from resulting in deeper gaps between those who control computing power, infrastructure, and data and those who depend on services produced in other countries or, in other words, he identifies the objective of inclusion but does not develop the instruments needed to achieve it.
He becomes much more concrete when he proposes using AI as a translation tool at the UN, pointing out that it could break linguistic blockages, reduce costs, and facilitate the participation of communities whose languages are not among the Organization’s six official languages. This orientation touches on a sensitive problem: translation algorithms carry epistemologies that impose certain visions when working on text, and they are not able to capture the subtleties of language, something that experienced human translators do achieve and that is essential in the diplomatic world. Sall does not address these limitations or explain how to guarantee quality, human control, and sovereignty over linguistic data.
Sall also refers to the need to establish a global cooperation platform, but he does not define its mandate, its membership, or its relationship with existing mechanisms, which marks a constant: much more concrete positions on AI in administrative matters than in the area of its international governance.
María Fernanda Espinosa: AI, Employment, and Multilateral Governance
In her vision document, she proposes combining the use of AI with the experience of staff to determine where political progress remains possible, and she proposes deepening the Global Digital Compact and the global dialogue on AI, underlining elements of democratization and expansion of the capacities to use AI by everyone as a cross-cutting element of her positioning.
During the dialogue, Espinosa linked technology with employment and the need for institutional coordination: in response to a question from Djibouti on how to mobilize the United Nations system around a global employment agenda, she said that “When I look at the impact of artificial intelligence, I strongly believe that we really need to be serious about aligning the UN system.”
She presents AI as a problem of multilateral governance and distributive justice: when she speaks of supporting AI “for the future, not the few,” she makes explicit her concern about the concentration of its benefits, even though her approach remains within a broad programmatic language: responsibility, inclusion, prosperity, and cooperation appear as the four central categories that politically orient her proposal, but without reaching specific proposals on how to redistribute capacities or what obligations should be imposed on the actors that control the systems.
Along those lines, she assigns to the United Nations the main function of being the convening and coordinating institution for debates at the global level, understanding that it should provide a space to discuss risks and opportunities, continue the implementation of the Global Digital Compact, and align the different entities of the system. However, the candidate is not precise about how to turn dialogue into commitments or prevent the formal participation of all countries from functioning as a veil that conceals the enormous differences in their effective negotiating capacities in the field of AI.
Despite that criticism, this position clearly seeks to make use of the UN’s real institutional advantage as the only forum with universal reach and general competences.
The candidate’s emphasis on the AI and employment axis adds a social dimension that avoids treating AI as a matter of innovation and puts broader social factors on the table: in a reference to the situation of African youth, she recognized that the consequences of automation will differ depending on the demographic, productive, and educational structure of each region.
Her position is more developed than that of several other candidacies, but it still does not translate its ideas into concrete actions aimed at establishing common rules, responsibilities of actors, or accountability mechanisms.
Carolyn Rodrigues-Birkett: Institutional Use of AI in the United Nations
In her vision statement, Rodrigues-Birkett does not develop AI as an autonomous priority within a document focused on conflict prevention, sustainable development, human rights, climate change, the situation of small States, and institutional reform.
During the dialogue, she does refer to the use of AI within the United Nations: responding to a question asked by Andorra, on behalf of the Group of Francophone Ambassadors, on multilingualism within the United Nations, she stated that “The Secretariat is already utilizing AI, with human review, and we have more work to do in making use of that tool.” The reference to human review appears as the most concrete normative component of her position, although, framed in overly general terms, it may be read as contradictory with other elements of her discourse, such as the application of AI to improve multilingualism at the UN, which we will see below.
Rodrigues-Birkett’s
position is, consequently, predominantly operational and lacks a general vision
of the transformation produced by AI at the global level or proposals for its
governance: the focus is clearly placed on the way the UN can use this technology
to better fulfill its functions.
However, the fact that she chose to refer to AI when asked about multilingualism makes applicable here what we said earlier regarding automated translations: problems of implicit biases and properly human capacities involved in that process remain unaddressed.
Rodrigues-Birkett’s position is cautious compared with that of other candidates who promote approaches with greater momentum for the use of AI in the UN, which makes its scope limited.
Olara Otunnu: An Institutional Proposal for AI
Artificial Intelligence, is fast emerging.”
His proposal to “ensure the creation of a special body to oversee the UN’s work on artificial intelligence, potentially including i) creating frameworks for the development, adoption, safety, governance, and societal and employment impact of AI, and ii) using AI and digital technology to help lift billions of people out of poverty through financial and digital inclusion” stands out.
The late presentation of Otunnu’s candidacy has meant that, as of the date on which this blog is being written (August 6, 2026), the official page of the selection process does not record an individual interactive dialogue between Otunnu and the General Assembly, so we can only say that he presents one of the most explicit and programmatically ambitious positions among the candidacies, in which AI is explicitly linked to production, employment, financial inclusion, the fight against poverty, and international security.
His vision statement combines developmental optimism with a perception of risks, but by focusing on extremes it loses sight of the more immediate governance problems, such as corporate concentration, the opacity of systems, algorithmic discrimination, dependence on private infrastructures, or the unequal distribution of capacities.
His consideration of financial and digital inclusion introduces a distributive dimension, but he does not explain through what mechanisms it could reach those who lack connectivity, technical capacities, or access to infrastructure, which creates a risk of “good intentions” that end up widening existing gaps.
Conclusion: More Diagnosis Than Governance Architecture
The analysis carried out shows that artificial intelligence occupies different places in the proposals of those seeking to lead the United Nations, grouped as follows:- Candidacies that incorporate AI as a structural transformation affecting development and different areas of social life and institutional functioning. Here we place Rebeca Grynspan, María Fernanda Espinosa, and Olara Otunnu.
- Candidacies that consider it in a limited, instrumental, or even marginal way. Rafael Grossi is located here, as the clearest case of marginal treatment, along with Carolyn Rodrigues-Birkett and Macky Sall.
These differences show how each candidate has integrated AI into their diagnosis of the contemporary international order and into their conception of the leadership that the Secretary-General should exercise.
The comparison also makes it possible to see a tension that runs through several proposals that promote greater use of AI within the United Nations, either generally or in certain sectors such as conflict prevention or automated translation, but none clearly refers to the safeguards needed for that adoption. As a result, key issues are left out of the debate, such as auditing, explainability, accountability, vendor selection, data protection, or the capacity for challenge by affected States or communities, to name just a few examples from a much longer list.
Nor does any proposal develop the dimension of power that is inherent to AI. There is talk of inclusion, reducing gaps, and access to the benefits of AI, but not of the concentration of computing power, data, talent, and infrastructure in a reduced group of companies and States.
As a result of both elements, most of the proposals identify risks, but they do not include mechanisms capable of modifying the effective distribution of capacities or containing the dangers. Otunnu’s proposal to create a specific body for addressing AI within the United Nations system remains far from that.
The vision statements and dialogues show that almost all candidacies recognize the problems and threats that come with AI, but also that there is a considerabledistance between recognizing the problem, something at which the UN hashistorically been very good, and formulating a strategy capable of governingit, which is what is truly complicated in a world with little trust in multilateralism, growing inequalities, and a backdrop of “digitality” that seeps into every decision, affecting equality in the capacity to decide even when working under fictions of democratic and inclusive participation.
If we take into account that the person selected to lead the UN will have, within the term of their mandate, the closing of the 2030 Agenda, and perhaps will be responsible for institutionally leading the process on the “post-2030,” that person will have to give the necessary impetus to the Pact for the Future, which includes the Global Digital Compact as an annex, and ensure that the UN modernizes, among other reasons, so that it can respond efficiently and effectively to a world that is already digital.
“The most impossible job in the world” is becoming ever more impossible, but there is something that AI should not change: the nondelegable responsibility of the Secretary-General to exercise political judgment and leadership, make decisions, and answer for them.
References
Espinosa Garcés, M. F. (2026). Vision statement: Candidate for Secretary-General of the United Nations. United Nations. https://www.un.org/pga/80/2026/05/12/letter-from-the-president-of-the-general-assembly-the-president-of-the-security-council-on-selection-and-appointment-of-sg-nomination-of-ms-maria-fernanda-espinosa-garces-for-the-position-of-secr/Grossi, R. M. (2025). Vision statement: Candidate for Secretary-General of the United Nations. United Nations. https://www.un.org/pga/80/2025/11/28/letter-from-the-president-of-the-general-assembly-on-secretary-general-nomination-candidate-rafael-mariano-grossi/
Grynspan Mayufis, R. (2026). Vision statement: Candidate for Secretary-General of the United Nations. United Nations. https://www.un.org/pga/80/2026/03/04/letter-from-the-president-of-the-general-assembly-the-president-of-the-security-council-on-nomination-of-ms-rebeca-grynspan-by-the-republic-of-costa-rica-for-the-position-of-secretary-general/
Otunnu, O. A. (2026). Outline agenda for the United Nations: Olara Otunnu’s vision statement. United Nations. https://www.un.org/pga/80/2026/07/27/letter-from-the-president-of-the-general-assembly-and-president-of-the-security-council-on-the-joint-letter-on-nomination-of-mr-olara-otunnu-by-the-republic-of-uganda-for-the-position-of-secretary-general/
Rodrigues-Birkett, C. (2026). Vision statement: Candidate for Secretary-General of the United Nations. United Nations. https://www.un.org/pga/80/2026/06/15/letter-from-the-president-of-the-general-assembly-the-president-of-the-security-council-on-selection-and-appointment-of-sg-on-nomination-of-ms-carolyn-rodrigues-birkett-for-the-position-of-secret/
Sall, M. (2026). Déclaration de vision: Candidat au poste de Secrétaire général de l’Organisation des Nations Unies. United Nations. https://www.un.org/pga/80/2026/03/03/letter-from-the-president-of-the-general-assembly-the-president-security-council-on-nomination-of-mr-macky-sall-by-the-republic-of-burundi-for-the-position-of-secretary-general/
United Nations. (2026). Selection and appointment of the next Secretary-General. https://www.un.org/en/sg-selection-and-appointment
United Nations. (2026, April 21). Rafael Mariano Grossi: Interactive dialogue for the position of the next Secretary-General of the United Nations [Video and automated transcript]. UN Web TV. https://transcripts.un.org/en/asset/k10/k105axdjuz
United Nations. (2026, April 22). Macky Sall: Interactive dialogue for the position of the next Secretary-General of the United Nations [Video and automated transcript]. UN Web TV. https://transcripts.un.org/en/asset/k18/k1846ni8aq
United Nations. (2026, April 22). Rebeca Grynspan Mayufis: Interactive dialogue for the position of the next Secretary-General of the United Nations [Video and automated transcript]. UN Web TV. https://transcripts.un.org/en/asset/k1g/k1g4kqhazq
United Nations. (2026, June 15). María Fernanda Espinosa Garcés: Interactive dialogue for the position of the next Secretary-General of the United Nations [Video and automated transcript]. UN Web TV. https://transcripts.un.org/en/asset/k1m/k1mxnayn7q
United Nations. (2026, June 18). Carolyn Rodrigues-Birkett: Interactive dialogue for the position of the next Secretary-General of the United Nations [Video and automated transcript]. UN Web TV. https://transcripts.un.org/en/asset/k1w/k1wnpi5m6v