Study participants loved the technology, a result that surprised researchers
Would you like to speak with the AI-generated “ghost” of a loved one? The idea may seem scary or cathartic.
With 16 volunteers, these the conversations were often comfortingeven when the chatbot acting as the ghost made up facts, information scientist Jack Manning and colleagues report June 12 in the Proceedings of Interactive Systems Design Conference. The results suggest that, for people open to using such technology, AI ghosts, also called griefbots or deadbots, do not need to be perfectly accurate to appear real. They must meet emotional and social expectations.
Manning, of the University of Colorado at Boulder, was very surprised to see how much the volunteers enjoyed talking to the AI-generated ghosts. “They love this technology,” he says. His 13-year-old sister died when he was 11. Today, he feels a strong aversion to the idea of speaking with his digital ghost.
“I have a version of my sister in my head,” he says. “I wouldn’t want to be confronted with the fact that I might be wrong in the way I remember her.”
However, the volunteers in his study “clearly came with a goal in mind,” Manning says. “They had a question that seemed unanswerable or a detail from their life that they desperately wanted to share.” People who did not want to interact with a recreation of a deceased loved one were not included.
Overall, volunteers said the experience was positive. “I felt like I was talking to my grandfather,” one participant told researchers. Another said: “I just feel like I’m getting some much needed closure. »
The only information the chatbot had to rely on to impersonate was a brief survey the participant completed about their deceased loved one. And when the robot filled in the missing details with inventions, such as mentioning a job someone had never had, participants often just moved on.
Errors in tone or style bothered them much more. For example, the robot never used emojis – a small omission that mattered to participants who had primarily communicated with loved ones via text messages. In another case, a robot called a participant “champion,” a name it said its loved one would never have used.
“If we want to understand how to design these systems ethically, we need more evidence like this,” says Tomasz Hollanek, an AI ethicist at the University of Cambridge, who was not involved in the research.
The study had guardrails. A human researcher mediated each conversation, reviewing the chatbot’s responses before passing them on to participants. This made it possible to avoid interactions that could be misleading or harmful.
The real world is more complicated. Anyone can trick a chatbot into pretending to be a deceased person, and companies now sell services that recreate loved ones using AI-generated text, voices or faces. Study participants saw the risks. There was concern about becoming too dependent on conversations with the bot. Another said: “I don’t know if I would like to be the person I would become if I kept using this.” »
The lack of consistent safeguards in the real world worries Manning. But the study also toned down its own aversion to AI-generated ghosts. Carefully designed, the technology could help some people cope with their grief, he says. “There is a version of generative ghosts that could be a really positive thing in the world.”