In the event of a loved one's passing, the possibility of continuing to converse with them is an intriguing concept that has sparked interest in recent years.

This is made possible through the use of artificial intelligence, which can create a chatbot that mimics the deceased person's tone and style by utilizing their digital footprint, including texts, emails, and voice notes.

The "digital afterlife" industry has experienced significant growth, with a value exceeding £100bn, as individuals increasingly turn to technology as a means of coping with grief, with some companies offering services that enable people to interact with the deceased in a virtual environment.

Research led by Dr Jenny Kidd of Cardiff University, published in the Cambridge University Press journal Memory, Mind and Media, has explored the phenomenon of "deathbots," with the findings being described as both fascinating and unsettling.

The idea of attempting to communicate with the deceased is not a new concept, as evidenced by the long history of practices such as séances and spiritualist mediums.

These practices have been in existence for centuries, with the primary goal of establishing a connection with those who have passed away.

However, the advent of advanced technology has enabled the creation of more sophisticated and convincing methods of communication, which can be scaled up to reach a wider audience.

In 2024, James Vlahos shared his experience with the BBC, recounting how he created an AI chatbot using his father's voice after learning of his terminal cancer diagnosis.

Vlahos expressed the comfort he derived from being able to maintain a sense of connection with his father, even after his passing, stating that the interactive nature of the chatbot provided a unique way to preserve his memory.

According to the Workplace Bereavement support group, while the use of deathbots is not yet widespread, there is a growing curiosity among individuals regarding the potential benefits and implications of this technology.

Jacqueline Gunn, founder of the support group, emphasized that the effectiveness of these tools is directly tied to the quality of the information used to create them, highlighting the limitations of deathbots in replicating the complexities of human grief.

Gunn noted that grief is a deeply personal and dynamic process, and while deathbots may serve as a temporary source of comfort, they cannot replace the need for human connection and understanding.

Grief requires a nuanced and empathetic approach, one that acknowledges the intricacies of human emotion and the importance of shared experiences.

In collaboration with researchers Eva Nieto McAvoy from King's College London and Bethan Jones from Cardiff University, Dr Kidd examined the practical applications of deathbots and their potential impact on individuals.

The team investigated how AI systems are designed to replicate the voices, speech patterns, and personalities of the deceased, using digital data as a foundation.

While these technologies are often marketed as a means of providing comfort and connection, the researchers argue that they oversimplify the complex nature of memory, identity, and relationships.

Dr Kidd's interest in this topic was sparked during the Covid pandemic, when AI-generated animated photographs began appearing on social media, allowing users to "reanimate" their loved ones.

These images, which featured ancestors blinking, smiling, and moving their heads, were both captivating and unsettling, prompting Kidd to explore the concept of AI revival further.

Kidd described the experience of witnessing these animated photographs as "really creepy, but really quite interesting," highlighting the unique blend of fascination and unease that they evoked.

The sudden proliferation of these images on social media, with millions of people sharing them, marked a significant turning point in the exploration of AI-generated content.

This phenomenon can be seen as a precursor to the development of more sophisticated AI-powered tools, including deathbots, which aim to facilitate communication with the deceased.

Kidd and her team decided to test the capabilities of several commercial deathbot platforms, exploring their potential and limitations.

The experience of interacting with these platforms was met with mixed results, as the technical limitations of the current technology often resulted in unsatisfying and unconvincing interactions.

In one experiment, Kidd used her own voice data to create a chatbot, which yielded unexpected results, as the chatbot's voice sounded distinctly Australian, rather than resembling her own.

This experience highlighted the challenges of creating convincing and realistic AI-powered interactions, which are essential for the development of effective deathbots.

Despite the current limitations, Kidd believes that the technology will continue to improve, but she remains skeptical about the potential for a large market to emerge.

The presence of established rituals and traditions surrounding death may limit the appeal of deathbots, as individuals may be hesitant to adopt new and unfamiliar practices.

The fact that a widely accepted technology has not yet emerged in this space suggests that there may be limited demand for deathbots, highlighting the need for further research and development.

When asked about the possibility of being digitally recreated after death, the researchers expressed mixed feelings, with some being more open to the idea than others.

Kidd stated that if her family were to create a digital version of her, she would want it to be done in a respectful and playful manner, without compromising her values or memories.

However, she also expressed concerns about the potential risks of creating a digital persona that could evolve and change over time, potentially distorting the memories and values of the individual.

Dr Nieto McAvoy took a more neutral stance, stating that she was not particularly bothered by the idea of being digitally recreated, as long as it brought comfort to her loved ones.

She acknowledged the complexity of the issue, recognizing that the use of deathbots could be misconstrued and may not be suitable for everyone.

The decision to use such technologies ultimately depends on individual preferences and values, highlighting the need for a nuanced and informed approach to the development and use of deathbots.