Related Experiment Video
Updated: Aug 5, 2026
11:49
Enhanced Genetic Analysis of Single Human Bioparticles Recovered by Simplified Micromanipulation from Forensic ‘Touch DNA’ Evidence
Published on: March 9, 2015
Artificial intelligence in forensic science: Evaluation of ChatGPT for post-mortem interval estimation and Henssge
Elena Giovannini1, Simone Santelli1, Jacopo Lenzi2
1Department of Medical and Surgical Sciences, Unit of Legal Medicine, University of Bologna, Italy.
Summary
Artificial intelligence (AI) shows promise for forensic science but struggles with practical post-mortem interval (PMI) estimation. While accurate on theory, ChatGPT failed real-world scenarios and nomogram calculations, indicating it
Area of Science:
- Forensic Science
- Artificial Intelligence
- Natural Language Processing
- Thanatochronology
Background:
- Large language models (LLMs) like ChatGPT are emerging AI tools.
- Healthcare applications of ChatGPT are increasingly studied.
- Forensic applications of AI, particularly ChatGPT, are under-explored.
Purpose of the Study:
- To evaluate ChatGPT's performance in estimating post-mortem interval (PMI).
- To assess AI's capability with thanatochronological changes and the Henssge nomogram.
Main Methods:
- Fifteen PMI estimation questions were posed to ChatGPT.
- Questions covered theoretical concepts, practical applications, and Henssge nomogram use.
- Responses were evaluated for focus, accuracy, and completeness.
Main Results:
- ChatGPT demonstrated high accuracy and completeness for theoretical PMI questions.
- The AI model failed to accurately respond to practical case scenarios.
- ChatGPT was unable to correctly calculate PMI using the Henssge nomogram.
Conclusions:
- AI, including ChatGPT, has potential in forensic science but faces limitations.
- Uncertain data sources and incomplete literature access can impact AI accuracy.
- Currently, AI is not suitable for practical PMI estimation in forensic contexts.
