# Governing with artificial intelligence Author: Ignacio Martín Granados Original publication: 2025-10-08 Spanish original: https://estrategiabyaleph.substack.com/p/estrategia-106-donde-pisa-fuerte English URL: https://elcontemplador.github.io/estrategia-english/essays/106/ Status: Published translation This is a translation of the original Spanish essay published on 8 October 2025. Its claims, examples and forecasts retain that historical context. English publication: 2026-09-29 > We are delighted to welcome **[Ignacio Martín Granados](https://www.linkedin.com/in/ignaciomartingranados/)** back as a guest author, [following his contribution to issue 102](https://open.substack.com/pub/estrategiabyaleph/p/estrategia-102-como-la-ia-esta-cambiando?r=2tyybn&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false). A political scientist, prominent and active member of the Political Communication Association (ACOP), and lecturer on various postgraduate programmes, Ignacio combines extensive experience in political and institutional communication with a clear emphasis on public management and measurable impact. > > In this issue, Ignacio Martín Granados takes the [OECD report](https://www.oecd.org/en/publications/governing-with-artificial-intelligence_795de142-en.html) *[Governing with AI](https://www.oecd.org/en/publications/governing-with-artificial-intelligence_795de142-en.html) (2025)* as his starting point to map practical action on AI in the public sector: where it creates most value, where gaps remain, and which enablers and safeguards allow it to scale with confidence. He identifies opportunities—automation, personalisation, forecasting and accountability—warns of risks and biases, and offers criteria for prioritising high-benefit, low-risk use cases: a key element in moving from pilots to public policy with impact. In recent years, governments around the world have made significant progress in digitising the public sector, accelerated by the challenges posed by the COVID-19 pandemic. The focus today is on using digital technologies to respond better to citizens' needs: providing efficient, personalised, accessible and inclusive public services that deliver more value for taxpayers' money, and improving communication and engagement with citizens. Artificial intelligence is one of the most transformative forces of the 21st century and is becoming an integral part of digital government worldwide. In this fast-moving digital landscape, learning from experience is the best way forward. The Organisation for Economic Co-operation and Development (OECD) has therefore recently published [*Governing with Artificial Intelligence*](https://www.oecd.org/en/publications/governing-with-artificial-intelligence_795de142-en.html) (*Governing with Artificial Intelligence. The State of Play and Way Forward in Core Government Functions*, OECD, 2025). The first report of its kind to examine how governments use and engage with AI, it analyses 200 use cases and dozens of policy initiatives to understand how AI is applied and promoted across 11 core areas of government. Its aim is to give governments the elements needed to use AI effectively and identify areas where it is already having an impact and where gaps remain. [![Infographic on the presence of AI in government functions and the challenges of implementation, translated below.](https://elcontemplador.github.io/estrategia-english/assets/images/ddb231f6-ddb8-4d3c-9585-10be906e777e_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!xqKE!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fddb231f6-ddb8-4d3c-9585-10be906e777e_1536x1024.png) *Infographic text, translated: “Governing with artificial intelligence”; “The current state and future prospects of core government functions”.* | Where is AI most present? | Implementation challenges | |---|---| | Public services, justice and citizen participation | Skills shortages | | Public procurement, financial management, fighting corruption, public integrity, design and implementation | Difficulties obtaining and sharing quality data | | Policy evaluation, tax administration and administrative reform | Lack of concrete guidance | | | Insufficient monitoring and evaluation | | | Financial cost | | | Obsolete technological infrastructure | *The two columns are independent lists; their rows do not imply paired relationships. The coloured bars have no numerical labels. The footer cites “Governing with Artificial Intelligence” and OECD; its final year digit is partly obscured.* ## How AI can serve citizens The findings suggest that, in terms of total use cases, AI is most prevalent in public services, justice and citizen participation, while it is used less in policy evaluation, tax administration and public administration reform. Between these two extremes lie public procurement, financial management, fighting corruption and promoting public integrity, and regulatory design and implementation. According to the report, possible explanations for this distribution include the fact that some functions cover a wider variety of uses, as with public services, while others are narrower, as with tax administration. Some also face more regulatory restrictions—for example, tax authorities' use of tax data—while others encounter fewer implementation barriers and can mature faster, as in citizen participation. In certain fields, such as the administration of justice, public demands and case backlogs may accelerate AI adoption as an opportunity to address urgent challenges. AI use is more common in internal operations and public service delivery, but less so in government oversight functions. It is also less visible in policymaking, according to previous OECD analyses. In terms of benefits, most cases aim to promote automated, simplified and personalised processes and services, followed by better decision-making and forecasting, and greater accountability and anomaly detection. ## Risks of using AI in government Starting from the premise that there is no risk-free AI adoption, unlocking its benefits requires mitigating its risks. Biased algorithms can produce adverse outcomes; misuse of AI can violate or prevent the free exercise of citizens' rights; a lack of transparency, explainability and public understanding of AI can erode accountability and provoke public resistance; and excessive reliance on AI can widen digital divides and allow systemic errors to spread, weakening citizens' trust in government. These risks may arise from public officials' misuse of AI and be amplified in countries that lack mechanisms to guarantee the protection and promotion of human rights. Governments' failure to adopt AI also represents a risk, since it means missing opportunities for benefits and widening the capability gap with the private sector. Governments will need to incorporate AI if they want to respond to growing public demands and strengthen public trust. Ignoring the AI transformation or waiting until every unknown has been resolved will relegate governments to passive recipients of technology, rather than shapers of the available choices, incurring significant costs and disadvantages. [![Infographic setting out the benefits and risks of governing with artificial intelligence.](https://elcontemplador.github.io/estrategia-english/assets/images/6bb22456-c250-4ea4-b9fe-3c64dc180a76_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!B01K!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6bb22456-c250-4ea4-b9fe-3c64dc180a76_1536x1024.png) *Infographic text, translated: “Governing with artificial intelligence”; “The current state and future prospects of core government functions”.* **Benefits shown:** - Automated, simplified and personalised processes and services. - Better decision-making and forecasting. - Greater accountability and anomaly detection. - Responding to growing public demands and strengthening public trust. **Risks shown:** - Biased algorithms → adverse outcomes. - Misuse of AI → preventing the exercise of citizens' rights. - Lack of transparency, explainability and public understanding of AI → erosion of accountability. - Excessive reliance on AI → digital divide. - Ignoring the transformation of AI and its adoption by governments → lost opportunities and a wider gap with the private sector. *Credit printed in the image: “Governing with Artificial Intelligence. OCDE, 2025”.* ## Implementation challenges The difficulty of scaling successful AI applications means that many government initiatives remain at the pilot stage. Since this is still an early phase, skills shortages and difficulties obtaining and sharing quality data can be observed across all government functions. Moreover, although government AI strategies are common, the lack of concrete guidance hinders implementation. These factors reinforce risk aversion, limiting governments' ability to innovate with AI. Insufficient monitoring and evaluation mechanisms also restrict their ability to measure progress, detect risks and demonstrate a return on investment. Financial costs are another common challenge. Nor should we forget that obsolete technological infrastructure constrains the use of AI in public management. ## Ensuring trustworthy AI use To benefit from AI in government, mitigate its risks and overcome implementation challenges, the OECD proposes that governments establish the following: - Enablers for trustworthy adoption, including governance, data, digital infrastructure, skills, financial investment, agile procurement processes and the capacity to collaborate with non-governmental actors. - Safeguards to guide AI use, including rules and policies, guidance frameworks, transparency and accountability mechanisms throughout AI systems' life cycles, and oversight and advisory bodies to guide and evaluate these efforts. - Engagement mechanisms that promote user-centred approaches and effective responses, including interaction with key stakeholders such as the public, civil society and businesses. Ultimately, [the OECD encourages governments](https://oecd.ai/en/gov) to prioritise high-benefit, low-risk AI applications wherever possible, especially in the early stages of maturity. Yet most lack processes for comprehensively measuring potential or achieved outcomes—spending efficiency, service quality and possible harms—to support informed decisions. Addressing this gap should be a priority to ensure that AI implementations are transparent, trustworthy, fair and safe, and to begin making use of their full potential. *[Ignacio Martín Granados](https://www.linkedin.com/in/ignaciomartingranados)* *Political scientist and member of the Political Communication Association (ACOP).*