On November 11, 2025, Scientific American published an insightful exploration into the emerging role of artificial intelligence in helping people cope with grief. The article, featuring science writer David Berreby’s personal encounter with so-called “griefbots,” sheds light on how AI technologies are transforming the grieving process, offering unexpected comfort, insight, and connection to those mourning the loss of loved ones.
When someone close passes away, the yearning for one last conversation is a deeply human experience. Many wish for a chance to clarify unresolved issues or express final words of love. While resurrecting the deceased remains beyond our reach, an increasing number of people are turning to generative AI tools such as Replika to simulate conversations with digital recreations of their lost loved ones. These griefbots are designed to capture the essence of a person through AI-generated dialogue that mimics their voice, mannerisms, and speech patterns.
David Berreby, who has written extensively on robotics and AI, embarked on a personal journey to better understand griefbots by trying them himself. His experience, and the stories of others who have used such AI companions, challenge common assumptions and fears around this technology. Berreby found that while many critics quickly condemn the idea of digitally re-creating deceased individuals as potentially harmful or unnatural, those who actually engage with griefbots often report a more nuanced and sometimes beneficial experience.
The process of creating a griefbot typically involves providing the AI with various personal materials—voice recordings, photographs, writings, or transcripts of conversations. Yet, these data alone are not enough to capture the full humanity of the person. Users must also supply their own impressions and memories, describing the personality traits, favorite topics, or unique quirks of their loved one. This personal input helps the AI generate responses that feel authentic and meaningful, blending factual information with the user’s emotional connection.
Berreby experimented with multiple griefbot services, finding that the experience is a delicate balance between technology and human memory. Importantly, the AI does not replace the real person but serves as a tool to assist in processing grief. This leads to a fundamental insight into how grief functions neurologically. Our brains struggle to reconcile the reality of loss with lingering neurochemical signals that suggest the person is still alive. Grief is essentially a gradual learning process in which the brain accepts the absence of a loved one over time.
In this context, AI griefbots can be seen as a novel kind of artifact or memory aid, similar to looking at photographs, listening to old recordings, or handling personal objects. They create an interactive experience that helps bridge the emotional gap during the difficult period when the bereaved are adjusting to loss. Unlike internal imagined conversations, griefbots provide externalized interaction through text or voice, making the process more tangible without being a radical departure from traditional ways people cope with grief.
One concern that arises with griefbots is the risk that such lifelike digital recreations might encourage people to cling too tightly to the past and delay moving on. However, research cited by Berreby suggests an unexpected outcome: individuals who interact with griefbots soon after losing someone are actually more likely to engage socially with others rather than withdrawing. The griefbot provides a nonjudgmental space where users can express their feelings and take the time they need without societal pressure to “get over” their grief quickly.
This finding highlights a broader cultural issue. Modern society often struggles to accommodate grief in a compassionate and open way. People in mourning commonly report feeling rushed or pressured to suppress their sorrow because of social discomfort with death and loss. Griefbots, in contrast, offer a space free from judgment or timelines, allowing people to work through their emotions at their own pace. In this way, AI may be fulfilling a need that society currently fails to meet—a safe and patient companion in times of profound sadness.
Berreby’s reflections suggest that the popularity of griefbots speaks to larger societal avoidance of death and dying. In a culture that tends to prioritize productivity and positivity, grief can be an unwelcome “downer.” The availability of AI companions that listen without impatience or discomfort may help fill the emotional void left by inadequate social support. Ideally, a more death-aware society would reduce the need for such artificial substitutes, but until then, griefbots serve an important role for many people.
Despite the concerns and ethical questions surrounding griefbots, Berreby concludes his exploration with cautious optimism. He acknowledges that while the technology must be handled thoughtfully—