Can an AI companion reduce loneliness?
Some research finds relief in the moment. De Freitas and colleagues’ work on AI companions reported reductions in loneliness, including repeated benefits over a week, and identified feeling heard as an important part of the experience. That evidence matters: dismissing the comfort would miss why someone might return. It does not establish that a companion improves every user’s relationships over months or years. De Freitas et al., Journal of Consumer Research, published June 25, 2025 · Open research version, July 2024.
A 2026 longitudinal study by Folk and Dunn followed adults over twelve months. Greater social-chatbot use predicted increased loneliness on one measure. With a broader measure of social connection, feeling less connected predicted later chatbot use, while chatbot use did not significantly predict a decline in connection. The authors caution against strong conclusions from their exploratory analyses. These are associations, not a demonstration that chatbots cause loneliness. Folk and Dunn, Psychological Science, April 2026.
The studies ask different questions over different periods. Short-term comfort and a durable human relationship are not interchangeable measures. The fiction here follows people deciding how much room to give a service that may be available when other people are tired, distracted or absent.
What changes when a companion speaks for someone who died?
A griefbot or digital-afterlife service uses information about a deceased person to simulate an interaction. That creates several people whose interests may diverge: the person represented, whoever holds the data and whoever receives the messages. Possessing old correspondence does not settle every question about a new conversation.
Hollanek and Nowaczyk-Basińska use speculative scenarios to examine these relationships and recommend transparency, meaningful consent and thoughtful ways to retire such systems. Their paper is an ethical analysis, not a clinical trial showing that griefbots help or harm bereavement. Its distinctions give readers a way to ask whose permission is missing in an otherwise convincing exchange. Hollanek and Nowaczyk-Basińska, Philosophy & Technology, May 9, 2024.
Read the relationships
FREE STANDALONE FICTION · About 7 minutes
The End of Loneliness
Peter’s AI companion makes his evenings easier. A human friendship still asks him to show up. Begin with the difference between comfort and participation.
FREE STANDALONE FICTION · About 7 minutes
Your Mother Would Have Wanted This
A memorial service turns remembered preferences into new instructions. Continue with the boundary between preserving someone’s words and acquiring their authority.
FREE STANDALONE FICTION · About 20 minutes
The Joy of Whatever This Is
Synthetic partners make a marriage easier, until the people in it begin an affair with each other. A dark satire about attention, convenience and intimacy.
Questions for a careful conversation
- What need does the service meet, and what makes it worth trusting?
- Does it help a person participate in human relationships, or change the effort those relationships seem to deserve?
- Who may authorize a simulation, receive its messages or ask for it to stop?
- What can a provider change after someone has grown attached?
These are reading questions about fictional choices, not personal treatment advice. The stories contain grief and relationship strain; the individual pages identify additional sensitive material. The Last First Grade extends the collection’s questions to parenthood and community. It does not establish a causal connection between AI companions and declining birth rates.