Talk, Listen, Connect: Navigating Empathy in Human-AI Interactions

Abstract: Social interactions promote well-being, yet challenges like geographic distance and mental health conditions can limit in-person engagement. Advances in AI agents are transferring communication, particularly in mental health, where AI chatbots provide accessible, non-judgmental support. However, a key challenge is how effectively these systems can express empathy, which is crucial in human-centered design. Current research highlights a gap in understanding how AI can authentically convey empathy, particularly as issues like anxiety, depression, and loneliness increase. Our research focuses on this gap by comparing empathy expression in human-human versus human-AI interactions. Using personal narratives and statistical analysis, we examine empathy levels elicited by humans and AI, including GPT-4o and fine-tuned versions of the model. This work aims to enhance the authenticity of AI-driven empathy, contributing to the future design of more reliable and effective mental health support systems that foster meaningful social interactions.