AI Tools in Public Defense: Research Presentation by Dominik Stammbach

AI is expected to reshape or automate work and offer solutions to pressing societal challenges. One such challenge is indigent defense, a constitutional guarantee in the United States. Yet public defenders are asked to do more with less: representing clients deserving of adequate counsel, while facing overwhelming caseloads and scarce resources. AI has been proposed as a remedy for this access-to-justice crisis, but its practical role and limits remain poorly understood. 

 

I will discuss opportunities and concerns around AI for indigent defense, based on insights derived from semi-structured interviews and an ongoing collaboration with the New Jersey Office of the Public Defender. In qualitative interviews, participants view AI as most useful for evidence investigation to analyze overwhelming amounts of digital records, with narrower roles in legal research & writing, and client communication. In collaboration with NJ OPD, we are developing the NJ BriefBank, a retrieval tool searching through internal briefs, designed to streamline legal writing and surface reusable arguments. We find a mismatch between academic benchmarks and the queries public defenders submit: Models trained on existing benchmarks often transfer poorly, while domain-specific data, legal domain adaptation, curated synthetic datasets and query expansion strategies improve retrieval. Our findings can inform future work on legal search, and outline a research agenda for practical AI applications for public defenders more broadly.