AI

AI and the Afterlife

As AI systems improve in both breadth and depth of performance, they make it possible to create increasingly realistic agents—including agents modeled on specific people. We anticipate that within our lifetimes it may become common practice for people to create “generative ghosts”: AI agents that represent the deceased and that the living can converse with.

These technologies are likely to reshape existing practices—and give rise to novel ones—around end-of-life planning, remembrance, and legacy, with profound legal, economic, emotional, and religious ramifications. This project examines how people understand and engage with AI afterlives, the design considerations they raise, and the benefits and risks of building agents that speak for those who have died.

Since naming that design space, we have run the first user-experience study of people in conversation with these systems, and a follow-up study is in data collection now. Using AI to help people explore personal, family, and cultural archives is a follow-up area we’ve been considering as well.

Researchers

Jack Manning, Dylan Thomas Doyle, Anthony Pinter, Daniel Sullivan, Joseph Trybala, Ella Krelovich, Jed Brubaker

Publications

  1. Designing Conversations with the Dead: How People Engage with Generative Ghosts Manning, Jack M. and Sullivan, Daniel and Doyle, Dylan Thomas and Pinter, Anthony T. and Brubaker, Jed R.
    Proceedings of the 2026 ACM Designing Interactive Systems Conference
  2. Speculating the Afterlife: Reimagining Possibilities for AI and Its Impact on Memorialization, Grief, and Continuing Bonds Brubaker, Jed R. and Borovica, Tamara and Gerber, Katrin and McGrane, Caitlin and Hjorth, Larissa
    Thanatos 14
  3. Generative Ghosts: Anticipating Benefits and Risks of AI Afterlives Morris, Meredith Ringel and Brubaker, Jed R.
    Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems
  4. AI and the Afterlife Brubaker, Jed R. and Morris, Meredith Ringel and Doyle, Dylan Thomas and Fiesler, Casey and Gibbs, Martin and McGrenere, Joanna
    Extended Abstracts of the CHI Conference on Human Factors in Computing Systems

Press