Responsible Management of Judicial Data in the Age of AI

31/03/2025

Julio Gabriel Mercado, an Argentine expert, is among the prominent participants in the 2024 edition of LACNIC’s Leaders program. This initiative offers financial support and mentorship for research on key Internet Governance (IG) issues that impact the region, particularly from the perspective of local and underrepresented communities.

Mercado is committed to promoting Open Justice policies that foster transparency, open data, accountability, collaboration, and public participation. As a consultant, he focuses on researching, analyzing information, and generating knowledge to help judicial institutions move toward these principles.

As part of the Líderes program, Mercado’s research project — “Better Data, Better AI:
Data Needs for Responsible AI in Justice” — aims to provide judicial institutions with a set of key principles for effective data governance. The project defines essential principles for managing judicial data, so it can be used ethically and effectively in building responsible AI systems. The broader aim is to integrate AI into transparent, collaborative, and people-centered digital justice processes—ensuring equitable benefits for all, especially those in vulnerable situations. By proposing principles for data governance and addressing the ethical challenges associated with AI, the project seeks to support the development of regulatory frameworks that not only foster innovation in the justice sector but also strengthen human rights and promote digital inclusion.

 What motivated you to apply for the Líderes program in 2024?

I work closely with institutional data in my day-to-day. While working at Argentina’s Ministry of Justice, I was involved in driving data transparency initiatives—both within the ministry and in partnership with various provincial judicial systems. It was a long learning journey, but it led to a major milestone: we created Argentina’s first open portal for judicial data.

I also lead the Open Data Working Group within the International Open Justice Network—an initiative coordinated by the Judicial Council of the City of Buenos Aires. This professional network brings together members from across the region who share a common interest in promoting “judicial openness” through data.

In 2023, I took part in the Internet Governance Diploma program at the Catholic University of Uruguay, held in Montevideo. There, I met members of the Líderes program who were participating in LACNIC’s Policy Shapers initiative. Connecting with them made me realize that Líderes aligned well with my profile. Although I work as a consultant, I see my role as going beyond just executing projects—I actively engage in research, writing, publishing, and teaching as part of my broader professional commitment. The combination of consulting practice and academic work—of hands-on experience and knowledge production—is, in my view, one of the core values that the Líderes program aims to foster. Without a fully academic profile, it was difficult to find the time and space to pursue this research project, even though it had been on my mind for quite a while. Thanks to the support of Líderes and LACNIC, I was finally able to take the pause I needed and dedicate the time this work truly deserved.

What were your main goals behind your research?

AI systems operate by analyzing large datasets to detect patterns through statistical modeling. This means that when I ask a question, the system generates the most statistically probable answer based on its training data.

This raises a key concern: the data used to train AI systems is not always of high quality.  And when the data is not reliable, the outcomes generated by AI may fall short as well. Despite how closely data and AI are intertwined, they are rarely considered as part of a single ecosystem.  We celebrate AI breakthroughs but give little attention to the quality and origin of the data driving them. Recent European regulations emphasize that, in order for AI systems to operate safely and in compliance with human rights standards, they must be trained on high-quality data that is properly managed. This brings us to the source of that data. Public institutions have been managing and publishing data for years, but often without a clear understanding of how what they share might influence the behavior of AI systems—a field that is still relatively new territory for many of them.

 What approach and methodology did you choose for the project?

My goal was to bridge two worlds: those who work primarily with data and those focused on AI. After reviewing hundreds of academic papers and reflecting on my own experience in open data initiatives, I selected a diverse group of interviewees from both fields.  I wanted to gain deeper insight into their viewpoints and identify areas where these two often separate fields could align.  For me, one of the biggest challenges was the overwhelming volume of information available on this topic, which made it difficult to pinpoint the most methodologically sound and reliable sources.

What were the most important findings?

My research aimed to identify the quality standards that should guide the publication of data. The key finding was that if we want AI systems to be truly responsible—that is, aligned with ethical and legal principles—then data governance must strike a balance across five essential criteria: standardization, ensuring data is published and managed in a consistent, structured way; accessibility, making data openly available so that everyone can access and use it; completeness, guaranteeing datasets are representative of all social groups, including those who are typically underrepresented; privacy, protecting sensitive or confidential information from exposure or misuse; cybersecurity, safeguarding data to prevent tampering, breaches, or any compromise to its integrity and security.

Why is this relevant to the LACNIC community?

There are clear points of connection, as the project addresses areas such as cybersecurity, privacy, and standardization—key technical considerations when aiming to bridge digital divides.

My goal with this research was to contribute to a more human-centered approach to digital transformation. Making that shift possible requires joint efforts across different actors, and the technical community is key to that change.

What impact did the program have on you, and why do you think others should take the opportunity to apply next year?

I would definitely suggest applying—there are not many spaces like this where you can reflect, connect, and build knowledge, particularly for people who are not dedicated to research full-time but still care deeply about these issues. Beyond the financial support, having the endorsement of a respected institution like LACNIC was incredibly valuable—it gave me the structure and credibility I needed to carry out a project I had been wanting to pursue for a long time. The program gave me a solid framework to finally take on a project I had been meaning to do for quite some time—and being part of Líderes also gave the work greater visibility. Another highlight was the chance to share ideas with other researchers from around the region. It was great to connect with people who are finding new ways to link different topics within Internet governance. LACNIC also pairs you with a mentor—an expert in the field. I worked with Julián Casasbuenas, director of a Colombian NGO focused on tech’s social impact, and his guidance really helped me organize my ideas.

I also created a shorter, more practical version of the research, which I presented at the IGF in Saudi Arabia. That version offered a broader and more hands-on perspective on the topic. On a personal level, the project also enriched my day-to-day work. It gave me stronger, clearer arguments to talk about data quality and helped me organize a lot of the material I had been reading over time.

Looking ahead, Mercado is planning to present the Líderes project at an upcoming event organized by the International Open Justice Network. He intends to further explore the intersection of AI and data governance, with a focus on real-world quality standards and the emerging debate around data sovereignty, a growing concern in the field.