Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploring the Effect of Robotic Embodiment and Empathetic Tone of LLMs on Empathy Elicitation

Liza Darwesh, Jaspreet Singh, Marin Marian, Eduard Alexa, Koen Hindriks, Kim Baraka

TL;DR

The study investigates whether robotic embodiment and an empathetic tone in LLM-driven social agents affect empathy elicitation toward a third party using the Katie Banks paradigm. In a between-subjects design with four conditions (robot vs chatbot, empathetic vs neutral prompts) and 60 participants, participants read a story and interacted with an agent, then reported volunteering hours and completed empathy and interaction questionnaires. The main finding is that neither embodiment nor empathetic tone produced significant differences in volunteering hours or perceived empathy, with some evidence that prior familiarity with the robot influenced interaction quality; the study notes LLM robustness and occasional hallucinations as challenges. The work highlights the complexity of eliciting genuine empathy in HRI and suggests refining evaluation metrics, interaction flows, and experimental power to better capture subtle effects.

Abstract

This study investigates the elicitation of empathy toward a third party through interaction with social agents. Participants engaged with either a physical robot or a voice-enabled chatbot, both driven by a large language model (LLM) programmed to exhibit either an empathetic tone or remain neutral. The interaction is focused on a fictional character, Katie Banks, who is in a challenging situation and in need of financial donations. The willingness to help Katie, measured by the number of hours participants were willing to volunteer, along with their perceptions of the agent, were assessed for 60 participants. Results indicate that neither robotic embodiment nor empathetic tone significantly influenced participants' willingness to volunteer. While the LLM effectively simulated human empathy, fostering genuine empathetic responses in participants proved challenging.

Exploring the Effect of Robotic Embodiment and Empathetic Tone of LLMs on Empathy Elicitation

TL;DR

The study investigates whether robotic embodiment and an empathetic tone in LLM-driven social agents affect empathy elicitation toward a third party using the Katie Banks paradigm. In a between-subjects design with four conditions (robot vs chatbot, empathetic vs neutral prompts) and 60 participants, participants read a story and interacted with an agent, then reported volunteering hours and completed empathy and interaction questionnaires. The main finding is that neither embodiment nor empathetic tone produced significant differences in volunteering hours or perceived empathy, with some evidence that prior familiarity with the robot influenced interaction quality; the study notes LLM robustness and occasional hallucinations as challenges. The work highlights the complexity of eliciting genuine empathy in HRI and suggests refining evaluation metrics, interaction flows, and experimental power to better capture subtle effects.

Abstract

This study investigates the elicitation of empathy toward a third party through interaction with social agents. Participants engaged with either a physical robot or a voice-enabled chatbot, both driven by a large language model (LLM) programmed to exhibit either an empathetic tone or remain neutral. The interaction is focused on a fictional character, Katie Banks, who is in a challenging situation and in need of financial donations. The willingness to help Katie, measured by the number of hours participants were willing to volunteer, along with their perceptions of the agent, were assessed for 60 participants. Results indicate that neither robotic embodiment nor empathetic tone significantly influenced participants' willingness to volunteer. While the LLM effectively simulated human empathy, fostering genuine empathetic responses in participants proved challenging.

Paper Structure

This paper contains 10 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Related work
  3. Methodology
  4. Study design
  5. Participants
  6. Procedure
  7. Interaction Design
  8. Technical implementation
  9. Results and discussion
  10. Conclusion

Figures (4)

  • Figure 1: Experimental setup with the robot, where the participant is seated in front of the robot during the interaction.
  • Figure 2: User interface of the chatbot, showing the initial message to every participant and a response generated by the LLM.
  • Figure 3: System overview illustrating the interaction design.
  • Figure 4: Average scores for each selected subconstruct from the Almere model heerink, including interaction, likability, perceived sociability, social presence and trust, assigned by participants to agents from each experimental condition.