In a groundbreaking study published in the International Journal of Legal Medicine, researchers have put ChatGPT, a leading AI language model developed by OpenAI, through the rigorous testing grounds of a specialist forensic pathology examination. The results, which integrate the cutting edge of artificial intelligence with the nuanced and often ambiguous field of forensic pathology, provide remarkable insights into the potential and limitations of AI in highly specialized medical and legal domains.
Forensic pathology is an expert-driven field that involves meticulous examination of deceased individuals to determine the cause, manner, and circumstances of death. This specialty demands a combination of sharp analytical skills, a profound understanding of human anatomy, and the ability to interpret complex medical and situational evidence. Traditionally, such tasks have been the exclusive remit of highly trained forensic pathologists, whose judgments carry consequential weight in judicial settings.
The team led by de Boer, Young, Bouwer, and colleagues sought to explore whether an advanced AI language model like ChatGPT could reliably replicate the decision-making process in forensic pathology examinations. They employed a standardized and specialist exam designed to test profound knowledge spanning pathology, toxicology, scene analysis, and medico-legal principles, all reflective of real-world forensic challenges.
The AI’s performance was not only assessed on its accuracy in identifying causes of death but also on its reasoning ability, interpretation of ambiguous information, and contextual understanding—a critical factor given forensic pathology’s often complex and multifactorial cases. This multidimensional evaluation presented a unique framework to measure AI capabilities beyond mere factual recall, emphasizing cognitive reasoning in a specialist domain.
Remarkably, ChatGPT’s performance on the forensic pathology examination highlighted both its substantial potential and clearly delineated boundaries. The model demonstrated an impressive command of general forensic concepts and exhibited reasoning patterns that aligned with expert interpretations in many standard case scenarios. This suggests that such AI systems hold promise for augmenting medical examiners’ workflows, especially in providing preliminary assessments or assisting non-specialists with foundational forensic knowledge.
However, the research also illuminated areas where AI fell short, particularly in cases requiring nuanced judgment calls influenced by contextual subtleties or tacit clinical experience. These limitations underscore the indispensable role of human forensic pathologists in interpreting complex, ambiguous data in medico-legal environments where precision and accountability are paramount.
The study’s authors discuss the broader implications of integrating advanced AI tools like ChatGPT into forensic practice. While the AI’s capacity to process vast amounts of textual and clinical data swiftly could revolutionize logistical aspects of forensic workflows, such as generating initial reports or cross-referencing literature, ethical concerns persist regarding reliance on AI in legal proceedings.
Concerns include potential misinterpretations of AI-generated conclusions by non-specialists, the black-box nature of AI reasoning, and the necessity to maintain chain-of-custody integrity and evidentiary standards. The research urges forensic institutions and legal frameworks to carefully evaluate the role of AI in augmenting, rather than replacing, expert forensic acumen.
Critically, this study also opens discourse on training and education. The authors propose that AI tools like ChatGPT could serve as invaluable adjuncts in forensic pathology education, offering trainees dynamic case simulations and instant feedback on complex diagnostic scenarios, thus accelerating learning curves.
Moreover, by democratising access to forensic knowledge, AI could empower medical professionals in resource-limited settings who lack specialist pathologists. This could enhance death investigation quality globally, ultimately contributing to better public health data and legal justice delivery.
From a technical perspective, the researchers leveraged specific prompt engineering techniques to coax detailed, structured responses from ChatGPT, simulating the logic of forensic reasoning. They observed that nuanced prompts improved AI output relevance and accuracy, emphasizing the importance of tailored AI-human interaction in specialist domains.
The investigation also highlights the potential for integrating ChatGPT with other AI modalities, such as computer vision models analyzing autopsy images or toxicology data, to create comprehensive forensic AI suites that synergize multiple data streams for enhanced diagnostic yield.
This pioneering research sets a precedent for similar studies across other medical-legal specialties, where the intersection of AI and domain expertise could transform both professional practice and judicial processes. As AI technology continues to evolve rapidly, dialogues between forensic scientists, technologists, ethicists, and legal authorities become ever more crucial.
In conclusion, de Boer and colleagues’ study serves as a seminal exploration into the capabilities and constraints of AI in the demanding forensic pathology landscape. It portrays an emergent future in which AI acts as both a catalyst and collaborator—empowering forensic experts, enhancing training, and supporting judicial outcomes—while recognizing that the final interpretative authority remains human.
The promise and perils outlined by this research demand thoughtful stewardship as forensic pathology navigates its AI-augmented future, ensuring that technological innovation complements, rather than compromises, the pursuit of truth in medico-legal investigations.
Subject of Research: The performance and implications of ChatGPT, an AI language model, on a specialized forensic pathology examination and its potential role in forensic pathology practice and education.
Article Title: ChatGPT’s performance on a specialist forensic pathology examination: implications for forensic pathologists and non-specialists.
Article References:
de Boer, H.H., Young, G., Bouwer, H. et al. ChatGPT’s performance on a specialist forensic pathology examination: implications for forensic pathologists and non-specialists. Int J Legal Med (2025). https://doi.org/10.1007/s00414-025-03699-5
Image Credits: AI Generated
DOI: https://doi.org/10.1007/s00414-025-03699-5
