<p>Understanding user experiences of ChatGPT is essential to measure the true value of this technological advancement. The majority of ChatGPT studies focused on its performance and capabilities by testing it on various domains. Researchers expressed their concerns about ChatGPT use that might undermine academic integrity. Little attention has been paid to studying ChatGPT user experiences that may improve human-chatbot interactions. This study leverages user comments from Reddit communities to capture public sentiments, major themes, and factors through a conversation lens. Sentiment analysis and thematic cluster analytics are utilized to achieve this aim. Our findings revealed that (1) public sentiments on ChatGPT use are mixed, but the public discourse is generally positive. (2) user (dis) satisfaction, user autonomy, and perceived system quality have impacts on user experiences. Emergent factors include perceived prompt quality, perceived response quality (e.g., human likeness), and AI hesitancy. (3) among the most discussed topics, ChatGPT users are fascinated with its capabilities, whilst others feel frustrated when they encounter network errors. Many users discussed censorship-related content filters. This study sheds light on the ethical and responsible adoption of ChatGPT. It provides a multi-dimensional landscape surrounding ChatGPT and unlocks new insights into the impact on a broader AI discourse.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Analyzing ChatGPT user experiences: a sentiment analysis and thematic cluster analytics approach

  • Shasha Teng,
  • Azril Bin Hamzah

摘要

Understanding user experiences of ChatGPT is essential to measure the true value of this technological advancement. The majority of ChatGPT studies focused on its performance and capabilities by testing it on various domains. Researchers expressed their concerns about ChatGPT use that might undermine academic integrity. Little attention has been paid to studying ChatGPT user experiences that may improve human-chatbot interactions. This study leverages user comments from Reddit communities to capture public sentiments, major themes, and factors through a conversation lens. Sentiment analysis and thematic cluster analytics are utilized to achieve this aim. Our findings revealed that (1) public sentiments on ChatGPT use are mixed, but the public discourse is generally positive. (2) user (dis) satisfaction, user autonomy, and perceived system quality have impacts on user experiences. Emergent factors include perceived prompt quality, perceived response quality (e.g., human likeness), and AI hesitancy. (3) among the most discussed topics, ChatGPT users are fascinated with its capabilities, whilst others feel frustrated when they encounter network errors. Many users discussed censorship-related content filters. This study sheds light on the ethical and responsible adoption of ChatGPT. It provides a multi-dimensional landscape surrounding ChatGPT and unlocks new insights into the impact on a broader AI discourse.