Letter: Loneliness and Artificial Companionship in Palliative Care

Dear Editor:

Loneliness in palliative care remains insufficiently examined in the context of artificial intelligence (AI). Recent discussions have highlighted the functional roles of AI technologies in this capacity and the risks of emotional substitution.^1,2 AI systems, including chat-based tools that simulate supportive dialogue with users, can create a sense of emotional presence and engage with loneliness. However, limited focus has been placed on how digital interactions can intensify loneliness. ² While loneliness is often defined as a subjective gap between desired and actual social relationships, ³ it may also involve a sense that one’s distress is inadequately recognized or shared with others. ⁴

When used without carefully considering their limitations or possible harms, AI companionship may amplify social withdrawal and influence how loneliness is experienced. ² In digital conversations, accelerated feedback and externally oriented attention can obscure the source of one’s distress, the space between what is felt and how others respond to it. Furthermore, during serious illness, individuals frequently rely on nonverbal communication, especially gaze and small gestures. This requires them to navigate the ambiguity arising from these exchanges and can cause anxiety about whether they have been heard and understood, a concern closely linked to experiences of loneliness. Pain and suffering in terminally ill individuals are often dynamic, uncertain, and difficult to describe or locate.^4,5 As a result, responses from others, including those mediated by AI,^1,2 tend to function based on assumptions rather than what is intended or meant, which can increase feelings of distress. Loneliness thus stems from a disruption of relationships rather than a technical problem of connectivity.

The concept of self-illness ambiguity ⁶ offers a useful clinical lens for understanding loneliness in serious illness, particularly when suffering becomes hard to communicate.^4,5 Self-illness ambiguity is a term used in psychiatry and phenomenology, describing the difficulty of discerning where personal agency ends and the consequences of illness begin, which may underlie loneliness in palliative contexts. Symptoms, emotions, and identity blur, leading individuals to hesitate over whether the source of distress belongs to the self or to the condition. For example, individuals may struggle to distinguish depression from illness burden. ⁶ Such uncertainty also affects interactions with others. Communication may falter when responses focus on what is perceived rather than what patients intend to convey. Over time, this can erode relational confidence, the expectation that one’s experience will be, at least partly, recognized by others. This erosion can intensify alienation and loneliness when neither the self nor others reliably acknowledge suffering. ⁶

Loneliness in palliative and terminal care may be closely tied to uncertainty in how individuals experience suffering and themselves in relation to others, including AI companions. Meanwhile, human encounters are also shaped by uncertainty and emotional complexity, from which meaning and understanding can emerge. Loneliness may not only deepen distress but also create conditions in which hope and purpose can arise. Meanwhile, the simplifying tendencies of AI companions risk flattening the very experiences that matter most. In this regard, human interactions are especially consequential. Human connection is not only felt but also lived by being held within another person’s concerns, which can, in turn, alleviate loneliness. ⁶ Such connection can inform individuals’ responses to illness and approaching mortality, inviting flexible attentiveness to uncertainty rather than relying solely on control-based approaches. Thus, care should involve not only actions but also presence, silence, recognition, and the capacity to remain with distress. This includes sustaining human bonds amid uncertainty in symptom progression, changes in self-experience, and evolving social communication, including in digital interactions. It also emphasizes the cautious use of AI companions, which may offer support ¹ but cannot fully replace the contextual, relational, and interpretive work involved in serious illness.

Author’ Contributions

S.T. wrote the first draft. S.T. and J.F. contributed to the article and approved the submitted version.