From Text to Touch: Choosing How Artificial Intelligence Communicates Empathy in Healthcare
Year: 2026
Type: article
Abstract: Artificial intelligence (AI) systems can offer patients empathic support in forms as different as a chatbot on a phone that fits in a pocket and a soft robot they can hug. Much research on empathy in AI, including large language models, focuses on text, which lacks the rich non-verbal cues conveyed through voice, visual expression and touch. This Perspective introduces a new conceptual framework, the embodiment–scalability spectrum, to help researchers weigh what each modality can offer patients against how widely it can be delivered. It orders communication channels by embodiment: how far each draws on the expressive repertoire of the human body. Voice adds prosody. Visual cues offer facial expression, gaze and gesture through virtual avatars or physical robots. Touch, the most embodied channel, can offer a hand to hold; no other modality can. But text and voice can reach patients through devices they already own, whereas supportive touch requires dedicated hardware wherever the interaction occurs. That hardware can demand substantial development, constraining both research and widespread use. Consequently, modality choice fundamentally shapes how patients experience empathy from an AI system and the scale at which that support can be delivered.
Language: English
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7224981280