AI for Abolition? A Participatory Design Approach

TL;DR

This paper addresses how AI, specifically large language models, can be aligned with abolitionist goals by engaging TJ/RJ practitioners in a participatory design process. It outlines a methodology grounded in participatory action research to co-create an evaluation framework and identify practical AI use cases that support restorative and transformative justice without reproducing oppressive power dynamics. The authors propose a value-reflective evaluation scheme and a constitution-like set of principles to assess LLM outputs, coupled with a plan to test resource-efficient models and gather practitioner-derived prompts. The work contributes a concrete process for value-aligned AI design in marginalized communities and outlines concrete next steps, including data collection, prompt engineering, and expert-in-the-loop annotation. Overall, the approach aims to democratize AI development, highlight data stewardship, and foreground abolitionist praxis in AI research and deployment.

Abstract

The abolitionist community faces challenges from both the carceral state and oppressive technologies which, by empowering the ruling class who have the resources to develop artificial intelligence (AI), serve to entrench societal inequities even more deeply. This paper presents a case study in participatory design with transformative and restorative justice practitioners with the goal of designing an AI system to support their work. By co-designing an evaluation framework for large language models with the practitioners, we hope to push back against the exclusionary status quo of AI and extent AI's potentiality to a historically marginalized community.

Figures (1)

• Figure 1: The full process to evaluate alignment between LLMs and TJ/RJ practitioner values