Harnessing artificial intelligence in criminal proceedings to aid lawyers in interview preparation

• In Event: POLWG Panel 28. Innovating Criminal Justice? AI and cultural sensitivity training for Interviews, Suspect Rights, and Intelligence-led policing

Abstract

The presentation examines how large language models—a type of artificial intelligence—can support interview planning and preparation in criminal proceedings, aiding lawyers in this complex task. Drawing on the investigative interviewing methodology, it focuses specifically on expert witnesses in the field of digital forensics. In practice, judges, defence attorneys, and prosecutors often lack sufficient expertise in digital forensics, which hinders their ability to properly evaluate expert opinions and conclusions. It undermines the lawyers' capacity to ask informed questions, follow up on complex details, and fully understand the expert’s explanations in an interview situation. Investigative interviewing, rooted in recognized human rights, modern science, and established best practices, is mainly used by law enforcement and related entities that conduct interviews and develop effective tactics. Although interviewing tactics based on the investigative interviewing methodology can be beneficial for expert witness interviews, their adoption by lawyers in Polish criminal proceedings remains practically nonexistent due to limited knowledge on this topic and the lack of training. The conclusions presented are based on a series of experimental trials with various free and paid large language models, each fine-tuned for this specific application.

Author

• Denis Solodov, University of Warmia and Mazury in Olsztyn