In 2024, James Vlahos told the BBC how he recorded his dad’s voice and created an AI chatbot after hearing the devastating news that he was dying from cancer.
He described how wonderful it was to keep a sense of his memory alive, and while it didn’t remove the pain of his death, he added: “I have this wonderful interactive compendium I can turn to.”
The Workplace Bereavement, external support group said it was not seeing widespread use of deathbots, but more curiosity from people.
“These deathbots and AI tools are only as good as the information they are given,” said founder Jacqueline Gunn.
“They don’t grow or adapt in the way grief does. For some they may offer a stepping stone, but they cannot be the destination.
“Grief is a deeply personal human response to death, needing time, understanding and human connection.”
Working with Eva Nieto McAvoy from King’s College London and Bethan Jones from Cardiff University, Kidd explored how these technologies function in practice.
They looked into how AI systems are designed to imitate the voices, speech patterns and personalities of people who have died, using their digital traces.
While they are often marketed as sources of comfort and connection, the researchers say they rely on a simplified understanding of memory, identity and relationships.