Can grief be dealt with through automation? Is digital immortality possible?
10 years ago, these questions would have been impossible to answer, but nowadays, modern scientists have managed to come close through the creation of griefbots.
A griefbot—also dubbed deadbot or deathbot—is a large language model built on available information about the deceased, such as social media, letters, photos, diaries and videos. One such model was Dr. Muhammad Aurangzeb Ahmad’s Grandpabot, an AI simulation created to act like his father, who passed away before meeting his granddaughters. Although this isn’t the first instance of AI simulating the deceased—Jason Rohrer’s Project December, for example—it is a first in its focus on providing the opportunity to get to know someone never before met.
Dr. Ahmad is a research scientist at the University of Washington’s Harborview Medical Center and an Affiliate Assistant Professor in the Department of Computer Science at the University of Washington Bothell Campus. To create Grandpabot, Ahmad used letters and audio recordings of conversations with his father, meticulously collected throughout his life. Though not recorded for the simulation at the time, the recordings of these interview-like conversations were the most valuable part of the project, he says.
Ahmad’s father, who died in 2013, never got to properly meet his grandchildren, as his youngest child was born two years later. Upset that his children wouldn’t get the chance to know a figure so important to him, Ahmad came up with the idea of an AI simulation and got to work. Though Ahmad finished the first version of it in 2016, he has continued updating it ever since. “My kids have grown up with this model,” he says.
However, what’s to stop this new creation from becoming a Black Mirror episode? Critics have long been concerned about the dangers of this kind of digital immortality: Could it oversimplify or even present unwanted memories of the deceased?
However, scholars have created ways to address the ethical components of these developing technologies. Law professors such as Edina Harbinja and Lilian Edwards standardized the concept of post-mortem privacy, while Assistant Professor Carl J. Öhman at the Oxford Internet Institute has conducted research on the management of digital remains.
Despite the excitement of his new innovation, Ahmad remains cautious. “The danger I foresee is that these companion bots can potentially be very addictive, and people may start neglecting real people and real relationships,” he says.
After all, the loneliness epidemic persists on a global scale, as people don’t form the amount of friendships as they did in the past. These bots offer easy comfort and connection without the effort needed to maintain true relationships, and that is what can lead to true social degradation over time, as Ahmad fears.
