Recently, ChatGPT has gained attention for its capability to provide clear and complex responses to human inquiries. Despite the widespread use of advanced AI chatbots like ChatGPT, there is still a considerable lack of research on the emotional qualities of ChatGPT’s responses, especially when compared to human responses across various fields. This study aims to explore the emotions conveyed in responses generated by ChatGPT and contrast them with the emotions conveyed in human responses, across two domains finance and medicine, utilizing RoBERTa and EmoBERT to sentiment analysis and detect emotions in the responses. Our findings reveal that both ChatGPT and human mostly showed neutral opinions in their responses. However, there were clear differences in certain fields. In finance, human responses generally exhibited more negative opinions than those of ChatGPT. In contrast, in the medical field, human responses were more positive compared to ChatGPT. In medicine, where queries frequently involve sadness and fear, human responses showed optimism, reflecting empathy and a positive viewpoint. ChatGPT’s responses, while optimistic, also included elements of sadness and fear. In finance, where questions frequently encompass emotions of anticipation and disgust, human responses mainly contain anticipation and optimism while ChatGPT’s responses also reflect these emotions. Additionally, keyword analysis showed that in medicine, ChatGPT’s responses are typically more factual and cautious, aimed at informing and educating, whereas human responses tend to be more emotionally supportive. On the other hand, human responses in the finance domain often suggest a more definitive and confident tone, in contrast, ChatGPT responses frequently present information in a cautious and non-committal manner, avoiding definitive predictions. This research provides valuable insights that could enhance ChatGPT’s design by incorporating greater emotional intelligence, potentially improving user experiences.