Abstract
OBJECTIVES: This article reviews the developments in artificial intelligence (AI) technologies and their current and prospective applications in endof-life communications. It uses Open AI's generative pre-trained transformer 3 (GPT-3) as a case study to understand the possibilities of AI-aided communication in Palliative Care. MATERIAL AND METHODS: Open AI's GPT-3 was taken as a case study where responses were generated through the GPT-3 beta playground (Davinci engine) and were scrutinised by six mental health professionals (MHPs) working in a palliative care setting in India. They were tasked to evaluate the responses generated by the AI (the identity was not revealed until a part of the study was completed) in a simulated palliative care conversation with another MHP posing as a patient. The aim was to undermine whether the professionals were able to detect that the responses were indeed generated by a machine and did they approve or disapprove of the responses. RESULTS: The GPT-3 playground with the right prompts produced remarkable, often surprising texts and responses that imitated human interaction. However, glitches such as redundancy were noticed along with strongly held opinions in certain questions related to faith, death, and life after death. CONCLUSION: AI-assisted communication in palliative care could be used to train professionals in the palliative care field using it as a simulation in training. It could also be used as a therapeutic intervention for the purpose of engagement and philosophical dialogue after certain modifications. However, it would have its own limitations such as it cannot replace a human agent just yet.