Scaling what works: AI adoption across Latin American and Caribbean governments

As part of the Government AI Campus, Apolitical is publishing a series of articles featuring leaders from the 2026 Government AI 100, a list recognising public servants around the world who are pioneering AI adoption, capacity building and regulation.

Apolitical spoke with Miguel Porrua, a senior leader at the Inter-American Development Bank (IDB), where he supports governments across Latin America and the Caribbean in designing and implementing digital transformation initiatives.

How do you think about measuring the success of AI initiatives?

At the Inter-American Development Bank, we work across 26 countries in Latin America and the Caribbean. We use a methodology focused on development impact, which guides all our activities. For us, AI is not the objective itself; it’s just a tool. Success is measured based on whether a project achieves its planned results. If it does, we deem it a success. If it doesn’t, we analyse why, learn and apply those lessons to the next project. We also listen to beneficiaries. Did the project help improve people’s lives, whether through better health services, security or other outcomes? If it did, based on results indicators, we scale it up.

What challenges do governments face when designing and implementing AI initiatives, and how can they address them?

We frame this through three main foundations: governance, data and infrastructure, and human talent.

On governance, there is a lack of institutional capacity and limited regulatory environments. Many countries do not yet have agencies responsible for designing and implementing AI policies.

On data and infrastructure, there are major gaps. Only around 30% of government procedures are conducted online, meaning much data is still paper-based. Connectivity is also a challenge, with around 40% of the population lacking access to fixed broadband. Computational capacity is very limited compared to other regions.

On human talent, there is a significant shortage of digital skills. The region produces relatively few specialists in data and related fields, which limits governments’ ability to implement AI initiatives.

To address these challenges, we support governments with training, policy development, infrastructure projects and data protection legislation.

How can governments embed fairness and inclusion into AI initiatives from the outset?

One approach is to follow established principles from organisations like the OECD and UNESCO, but principles alone are not enough. Governments need mechanisms to monitor and enforce compliance, such as algorithm audits, certification processes and sandboxes where solutions can be tested before deployment. Institutions must also have the capacity to oversee and enforce these standards, particularly in sensitive areas like health and justice.

At the same time, it’s important to support the private sector. For example, tools can help developers assess whether their AI systems comply with fairness and inclusion principles. So it’s about balance: oversight and monitoring alongside support for innovation.

Can you share an example of AI creating public value while respecting rights?

In Colombia, AI has been used by the government’s legal defence agency to manage tens of thousands of cases each year. AI tools help decide which cases to prioritise, which to negotiate and how to shape legal strategy. Over eight years, this has saved close to 1% of GDP annually.

In Brazil, in the state of Ceará, AI has been used in the judicial system to reduce case backlogs. Processing times were cut roughly in half, and system performance improved by about 45% in two years.

These examples show how AI can deliver efficiency while supporting better decision-making.

What should public servants understand to use AI responsibly?

Public servants need to be well-trained. They should understand how AI works, including how data quality and training affect outcomes. They should also be aware of limitations, such as the lack of local context that many tools have.

“We think that AI tools are a great, valuable mechanism to improve the human capacity, to augment the human capacity, not to substitute it.”

Humans must remain in the loop at all times, especially in high-stakes areas, like health or justice, where mistakes can have serious consequences. AI should be used as a complement to human decision-making, supported by clear policies and guidance.

How do you expect AI in government to evolve over the next few years?

We expect more countries to develop national AI agendas. Currently, only a small number have clear strategies. AI will likely expand in areas such as financial management, health, security and customs. It will also enable more predictive governments that can anticipate citizen needs and provide more personalised public services.

We also expect AI to help address inclusion challenges, for example, by enabling services in multiple languages, including Indigenous languages. Governments will need to use AI to manage growing populations with limited resources.

What excites you most about AI in government, and what concerns you?

What excites me most is the potential impact on health, natural disaster management and environmental protection. AI could accelerate medical breakthroughs and help prevent or manage disasters. It also has the potential to make digital services more inclusive.

Concerns include data privacy, the digital divide, labour market impacts, misinformation and unclear accountability when AI systems fail. Strong policies are needed to address these risks while enabling the benefits.