A doctoral candidate at the University of Colorado Boulder, Jack Manning, is studying the use of large language models to create "generative ghosts"—AI chatbots trained on a dead person's social media posts, emails, audio and video recordings, photos and other data. Manning, who lost his sister more than 10 years ago, said "it is definitely a growing sector" as "interest in how AI and grief intertwine" increases. Startups including Séance AI, You, Only Virtual and re;memory offer these services; re;memory lets users create lifelike avatars based on photos and voice recordings. In a June paper published in the Association for Computing Machinery, lead researcher Jed Brubaker, Manning and colleagues observed participants' reactions to AI ghosts of people they have lost. Manning explained that participants "were fascinated by how much an LLM can do with so little information." According to Manning, a simple "death bot" plays back verbatim statements but cannot simulate the deceased's manner of communication, while generative ghosts recreate the deceased's voice and manner of communication, allowing interaction.