Letter : Preliminary Ethical Considerations Regarding the Use of Empathic Artificial Intelligence in Individuals Desiring to Die Alone

As a medical ethicist and palliative care physician, I argue from an ethical standpoint that we should utilize empathic artificial intelligence (AI) to provide benefits to those who wish to die alone.

In her works, ¹ sociologist Glenys Caswell states that “for some people in certain circumstances, to die alone is a choice and represents, for them, a good death,” “the group for whom dying alone is least acceptable are those who are the survivors of any death: their family and friend…”

These are brilliant insights. However, the descriptive, nonnormative findings do not fully indicate the specific support that would be beneficial to those who wish to die alone.

Psychologists Paul Bloom and Michael Inzlicht praised the general use of empathic AI almost unconditionally. They even said that “if more people feel heard and cared for with the assistance of AI, this could increase human flourishing.”^2,3 In a more recent interview, Bloom stated, “It’s fine, we agreed to give opiates to dying nonagenarians…no one wants to withhold AI friends from elderly patients with dementia.” ⁴ If Bloom’s argument is plausible, expanding the use of empathic AI to include those wishing to die alone aligns with the principle of beneficence.

Caswell offered careful considerations regarding the resource shortages in palliative care settings. Once developed, AI offers exceptionally high cost-effectiveness. A resource allocation that utilizes highly efficient resources is also correct from a utilitarian standpoint.

Users have the right to know that the source is an AI that lacks real emotions. ² As a fundamental premise, the principle of respect for autonomy must be upheld. Furthermore, the development of a more capable AI is the social responsibility of AI researchers.

One concern is that becoming too accustomed to AI empathy might worsen isolation and loneliness. ² Therefore, caution is needed based on the principle of nonmaleficence. However, I believe this concern to be unfounded. Because these users will die soon, chronic and long-term use is unlikely.

If the opinions and values of the user’s close family and friends, as well as the user’s personal values and experiences, are inputted, the AI will become more customized and its empathy will likely be more effective.

When considering sociocultural factors that could influence grief care, particularly family and religion, palliative care could be accepted by different cultures. If empathic AI can be developed to be receptive to differences in culture, race, and religion, it will benefit the welfare of those who wish to die alone even further.

Empathic AI can be used in grief care for the bereaved. AI that is programmed with the happy memories of the deceased is likely to be useful, at least during the acute phase of bereavement. However, intuitively, long-term use may disable the psychological defense mechanism of human “forgetting.”

Finally, I assert that the use of empathic AI for those desiring to die alone has significantly more benefits than drawbacks. From an ethical standpoint, it is advocated that the sound application of AI is the right thing to do.