What Is Required for Empathic AI? It Depends, and Why That Matters for AI Developers and Users

TL;DR

This paper addresses the lack of consensus on 'empathic AI' by proposing a framework that treats empathic capabilities as constellation-driven rather than defining empathy as a single property. It identifies three medical AI use cases—medical question-answerers, AI care assistants, and AI care providers—and argues that each requires different combinations of empathy-related abilities. The authors introduce the concept of 'fine cuts' of empathy (including cognitive, affective, and motivated components, mirroring, information types, consciousness, accuracy, self–other differentiation, motivation, interaction, and vulnerability) to guide design and evaluation, supported by the idea that not all components are necessary for every application. They discuss practical implications for AI developers and users, emphasizing component-based design, transparency via model cards, ethical considerations, and the need for robust assessment methods and empirical research to determine which capabilities are desirable in which contexts.

Abstract

Interest is growing in artificial empathy, but so is confusion about what artificial empathy is or needs to be. This confusion makes it challenging to navigate the technical and ethical issues that accompany empathic AI development. Here, we outline a framework for thinking about empathic AI based on the premise that different constellations of capabilities associated with empathy are important for different empathic AI applications. We describe distinctions of capabilities that we argue belong under the empathy umbrella, and show how three medical empathic AI use cases require different sets of these capabilities. We conclude by discussing why appreciation of the diverse capabilities under the empathy umbrella is important for both AI creators and users.