Abstract
- This work explores the transformative potential of digital twins in the context of police investigations in a specific case, focusing on their intersection with Large Language Models (LLMs), Building Information Modelling (BIM) technologies and forensic crime scene analysis. LLMs represent a new class of cognitive tools that can semantically support complex investigative processes through contextual language processing, from structured analysis of unstructured data and information to hypothesis generation in dialogue-based interactions. In combination with digital twins of crime scenes, LLMs enable multidimensional, simulation-based, hypothesis-generated reconstruction of criminal events. For example, perceptual errors of witnesses can be systematically evaluated by comparing their statements systematically and situationally with digitized crime scene models. The integration of generative AI extends these approaches to model adaptive scenarios, reconstruct variants and create personalized training environments. The use of digital twins as dynamic, interactive investigative tools is the focus of the research question and opens a bidirectional connection between real and virtual information, which can be particularly valuable for complex analyses.