University of Otago health sciences education adviser Dr Jon Cornwall said AI-generated representations of dead people — thanabots — could soon replace human bodies in dissection rooms, and be used to support students’ anatomy learning in the future.
Thanabots (derived from thanatology, the study of death) are AI-powered, digital replicas of deceased individuals which are already being used to assist people during bereavement.
They are virtual avatars designed to simulate conversation with a deceased person, and are created by training large language models (LLMs) using the deceased’s digital footprints — such as text messages, emails, social media posts and photos — to mimic their tone, vocabulary and personality.
These systems — also known as ‘‘griefbots’’, ‘‘deathbots’’ or ‘‘digital ghosts’’ — are transitioning from speculative fiction into real-world applications.
Technologies such as Project December, which simulates text-based conversations with the dead, and Deep Nostalgia, which animates old photos, show how digital afterlives are increasingly represented and even normalised.
‘‘Use of griefbots for this purpose is controversial,’’ Dr Cornwall said.
‘‘There is a lack of strong evidence for benefit and arguments include substantial psychological harms.’’
While the technology is not being used in anatomy dissection rooms yet, Dr Cornwall and Harvard Medical School Associate Prof Sabine Hildebrandt said it may be just around the corner.
There, they would be called ‘‘thanatobots’’.
Their research showed thanatobots could hypothetically be loaded with a donor’s medical records, and could contain anatomy education software.
‘‘If used, potential benefits might include enhancing learning via linking medical history to anatomical findings and clinical reasoning through personalised teaching assistance,’’ Dr Cornwall said.
‘‘They may support professional development by facilitating humanistic engagement and illuminating donors as people rather than objects.’’
However, the research cautioned against their use at this stage.
He said it could produce significant risks, including known adverse psychological effects for students.
Emotional engagement with a digitally ‘‘resurrected’’ donor could overwhelm learners or cause unhealthy parasocial attachments.
The illusion of a human presence risks trivialising the body donor’s physical reality, and could compromise the learners’ authentic encounter with mortality and respect for the deceased.
Cultural norms and individual grief may be disrupted, especially for students already sensitive to exposure to the dead, or from backgrounds with strong constraints around postmortem representation.
This includes instances where death and the dead are considered sacred, and further engagement with their likeness is considered taboo.
‘‘Thanatobots may inadvertently reduce donors to inauthentic digital proxies, paradoxically harming humanistic engagement, and negatively impacting learning outcomes.’’
He said AI errors associated with generated content may also hinder accurate learning.
Further issues included identifying appropriate operational thanatobot parameters, ethical data management, concerns around appropriate consent, and a lack of legislative oversight.
He said any responsible educational use of thanatobots must carefully explore their psychological and social impacts.
‘‘Ethical, cultural, and pedagogical challenges around thanatobot development should be critically examined prior to any incorporation into anatomy education.’’
