Abstract
This paper introduces a novel perspective on empathy of generative AI systems (GAIS) and develops a Stimuli-Organism-Response based theoretical framework to enhance conversational GAIS. Regardless of its broader applications, this framework aims to equip GAIS with the ability to recognize and respond empathically to the emotions of information seekers. Employing machine learning and Natural Language Processing, EmoTune is being developed as a Proof of Concept to detect emotions in user inputs and align them with appropriate response strategies before engaging a pre-trained AI, such as ChatGPT, for empathetic response generation. EmoTune will subsequently assess emotions in AI responses before conveying them to users. The novelty of EmoTune is the introduction of an additional layer to the existing conversational AI architectures, recognizing and responding to users' emotions, thus enhancing user trust, satisfaction, and acceptance. Our paper opens new research opportunities for designing empathetic GAIS and evaluating their effects on users.
Recommended Citation
Nguyen, Lemai; Vu, Quan; Xiong, Bingqing Dr.; and Sharma, Rajeev, "Empathetic Information Seeking Support using Generative Artificial Intelligence" (2024). Proceedings of the 2024 Pre-ICIS SIGDSA Symposium. 25.
https://aisel.aisnet.org/sigdsa2024/25