Artificial Intelligence (AI) is now stepping into one of the most sensitive areas of human life, death and memory. A growing number of companies are developing 'deathbots', AI-powered chatbots that can mimic the voices, language, and personalities of people who have died. These systems promise to let loved ones 'speak' with the deceased, but they also raise serious questions about technology, emotion and ethics, reports The Conversation.

Add Asianet Newsable as a Preferred Source

Talking to the dead through data

Researchers behind a recent study published in Memory, Mind & Media decided to explore what happens when remembering the dead is handed over to algorithms. They even went one step further, by creating digital versions of themselves to see how these systems really work.

The study is part of a larger project called Synthetic Pasts, which examines how technology shapes both personal and collective memory. It focused on how AI systems claim to preserve or recreate a person's voice, memories, or digital identity after death.

What are 'deathbots'?

'Deathbots' are AI systems designed to imitate a person’s personality after they die. They use data such as emails, social media posts, text messages, and voice recordings to build an interactive avatar that talks and responds like the deceased person.

These AI replicas can simulate speech patterns, tone, and even humour, giving the impression that someone who has passed away is still communicating. Some companies even promote this experience playfully, inviting users to 'host a séance with AI'.

But for the people who actually try them, the experience can be emotional and unsettling.

How researchers tested deathbots

To understand these systems deeply, the researchers turned themselves into test subjects. They uploaded their own videos, voice notes, and messages to create ‘digital doubles’, AI avatars based on their personalities and data.

In one scenario, they acted as people preparing their own digital afterlife, recording stories, memories, and advice. In another, they became the bereaved, talking to an AI version of someone who had 'died'.

What they found was both fascinating and disturbing. Some systems focused on memory preservation, helping users record personal stories arranged by themes like family, childhood, or advice. The AI then acted like a digital guide, helping loved ones browse these memories later.

Others went further, using generative AI, technology that allows the bot to chat freely, using data about the deceased to create evolving conversations that seem 'real'.

When comfort feels artificial

While some users found comfort in hearing a familiar 'voice', the researchers found that the emotional tone often felt strange or out of place. The AI sometimes repeated phrases word-for-word from earlier conversations, making replies sound stiff and mechanical.

In one example, when discussing death, the bot replied with cheerful emojis and upbeat language, saying:

Such replies showed how AI struggles to handle complex emotions like grief and loss. Instead of offering empathy, it produced awkward responses that reminded users of the system’s artificial nature.

Even when the systems used real recordings and genuine memories, they often placed stories into rigid categories, losing the depth and nuance that make human memory rich and meaningful.

What the conversations revealed

One researcher wrote about a conversation between a human and a deathbot:

The response, though well-structured, felt emotionally flat, showing how 'synthetic intimacy' can feel hollow. The researchers concluded that these digital afterlives reveal more about the companies that create them than about the people they are supposed to represent.

The business of remembrance

Behind every deathbot is a business model. These platforms are not memorial charities, they are commercial ventures that charge subscription fees or offer 'premium' options.

Some companies even partner with insurers or health services, offering people the chance to 'capture their story forever' before they die. But these platforms also collect emotional and biometric data, turning memory into a product that can be measured, analysed, and monetised.

As technology scholars Carl Öhman and Luciano Floridi argue, the 'digital afterlife industry' has created a new political economy of death, where a person’s data keeps generating profit long after they are gone.

This business model is part of what Professor Andrew McStay calls the 'emotional AI economy', where human emotions themselves become valuable data.

The promise and the illusion of digital resurrection

Many companies promote the idea of a 'digital resurrection' bringing someone back through their data. These systems use voices, gestures, and memories to create real-time conversations that feel alive.

But researchers warn that this promise is misleading. AI can simulate empathy, but it cannot truly feel or understand it. The digital avatars may 'sound' real, but they operate strictly within the boundaries of their code.

As technology scholar Wendy Chun points out, digital systems often confuse 'storage' with 'memory'. They promise perfect recall but forget that true memory includes the ability to forget, something essential to both mourning and healing.

The limits of algorithmic empathy

AI systems that attempt to 'recreate' the dead risk misunderstanding death itself. Real death has finality, an end that allows life and memory to move forward.

When these systems make the dead seem always available, interactive, and updatable, they change how we deal with grief. Loss becomes endless, replayable, and data-driven.

The study concludes that while AI can preserve stories and voices, it cannot recreate the living essence of a person or the complex emotions of human relationships.

The 'synthetic afterlives' created by AI are fascinating precisely because they fail, they remind us of what can never truly be digitised: the depth of human experience.

Remembering what is real

In the end, the researchers found that while AI may allow us to 'talk' to the dead, what we hear back is not the voice of the person we lost, it is the reflection of our own data, shaped and sold by technology.

As one of the researchers put it, "These systems tell us more about ourselves, and about the platforms that profit from our memories, than about the ghosts they claim to let us talk to."