Digital Immortality Explained: Ideas, Products and Limits

1What does “digital immortality” mean?

The term describes preserving some part of a person in digital form so that it can outlast their body. That “part” can be very small, such as a collection of photos, or very ambitious, such as a complete copy of a mind.

The word “immortality” is misleading. No current technology keeps a person alive, conscious or aware. What digital tools can do is store information about someone, or imitate the way they spoke or wrote. Keeping this distinction in mind makes it much easier to judge products and headlines.

2Where the term comes from

An early and influential framing came from two Microsoft researchers, Gordon Bell and Jim Gray. In a paper titled “Digital Immortality”, dated October 2000, they described a spectrum. At one end is one-way immortality: preserving your ideas so that future generations can read them. At the other end is two-way immortality: a digital version that can respond and keep learning. They predicted that the two-way kind would become possible within this century.

Bell then became the test subject of MyLifeBits, a Microsoft Research project that tried to record and store as much of his life as possible, including documents, photos, calls and images from a wearable camera. In 2014 a start-up called Eterni.me proposed avatars that would learn from a person’s digital footprint and talk to relatives after death. Researchers at the University of Melbourne noted at the time that it had not yet produced a working avatar.

In 2017, Oxford researchers Carl Öhman and Luciano Floridi gave the commercial side a name: the digital afterlife industry. Their paper argued that companies handling the data of dead people have economic incentives that need ethical rules.

3The three approaches

1. Archives and memorials

This is the oldest and most common form. It includes photo collections, recorded life stories, letters, memorial pages on social networks and the “legacy contact” settings that let a trusted person manage an account after death. Nothing new is generated. The record simply survives, which is also its strength: everything in it is real.

2. Interactive AI simulations

These systems let you ask questions and get answers in the person’s style. There are two families. Recorded-answer systems play back real video or audio answers that the person gave while alive, so they never invent anything but can only answer what they were asked. Generative simulations use language models to write new replies based on a person’s messages or writing. They can talk about almost anything, but they can also invent words the person never said. Researchers often call simulations of the dead “griefbots” or “deadbots”; our guide to griefbots covers that research.

3. Mind uploading

Mind uploading, also called whole brain emulation, is the idea of scanning a brain in enough detail to run it as software. A 2008 roadmap by Anders Sandberg and Nick Bostrom at Oxford set out what this would require. Scientists have since mapped the complete wiring of an adult fruit fly brain and small samples of mouse and human brain, but nobody has emulated any brain in full. Our article on mind uploading versus AI clones explains how far the science has got.

4Overview: what exists and what does not

Science fiction often blends these categories. A film may show a simulation that has the person’s memories, feelings and awareness all at once. In reality, an archive has no awareness, a simulation imitates patterns in data, and a true upload does not exist.

5Limits that every approach shares

• It is not the person. Archives keep traces and simulations imitate them. Neither has experiences, feelings or awareness.
• Consent matters. Most people never said whether they would want to be simulated, and their messages also contain other people’s words.
• Services can close. Companies change or fail. StoryFile, a well-known recorded-video company, filed for Chapter 11 bankruptcy protection in 2024. Keep your own copies of anything important.
• Generated words can mislead. A generative simulation can produce confident statements the person never made.
• Grief is sensitive. In 2024, Cambridge researchers Tomasz Hollanek and Katarzyna Nowaczyk-Basińska called for safeguards such as clear AI labelling, consent and easy ways to retire a simulation.

Some people hope that one of these routes might lead to a longer life itself. Our honest answer to whether you can live forever with AI separates medical research from digital memory.

6Where Memory Clone fits, honestly

Memory Clone belongs to the second group, and it is a modest example of it. It is a Windows app, currently in alpha, that builds a private AI simulation from chats you export yourself (WhatsApp, Messenger, Instagram, Discord) and notes you write. It learns writing style separately for each relationship and can search dated memories in a local index.

• It preserves how someone wrote. It does not preserve the person, and every clone is presented as an AI simulation.
• It is text only today. It does not use voice notes, photos or video, and it does not clone voices.
• Imported originals stay on your PC. Cloud processing is off until you switch it on for a clone.
• You must be 18 or older, and cloning another living person requires their permission under our Acceptable Use Policy.

You can read how replies are generated on our AI transparency page, or see the download page if you want to try it.