Tine Destrooper

In the past two decades, critical voices have come to dominate scholarly debates over transitional justice. These voices have contributed to a number of important turns within the field: an actor-oriented turn, a local turn, a decolonial turn, a transformative turn, etc. Each of these turns has made invaluable contributions to the field, making the debate – and arguably the practice – more victim-oriented, context-sensitive, concerned with root causes of conflict, and self-reflective.

From within each of these normative orientations concerns can be articulated about the emergence of artificial intelligence (AI) and about its nexus with transitional justice. This nexus lies both in the use of AI as a research tool within the field of transitional justice, as well as in the emergence of AI as a practical tool for planning interventions. There is a booming literature on the risks and challenges of AI for transitional justice, for example related to biases, risks of undoing the actor-oriented trend, overlooking contextual knowledge and insights, or replicating colonial hierarchies.

Yet, as a recent report by Impunity Watch (2025) shows, there is also great – and mostly untapped potential – in terms of how AI and notably machine learning methods may be used to facilitate analyses that were hitherto impossible.

The project is rooted in critical transitional justice studies and considers which conditions need to be met for integrating ML-methods in empirical transitional justice scholarships, notably to use these methods in ways that aligns with transitional justice’s normative orientations of more just and inclusive societies.

Based on a test case that uses existing databases, it explores challenges as well as opportunities within and beyond transitional justice and human rights research.