When Artificial Intelligence Meets Human Intelligence: Topics and Sentiments about ChatGPT in Online Knowledge Sharing Communities

Abstract

OpenAI’s release of ChatGPT has sparked worldwide discussion and debates about its potential implications. Knowledge sharing conversations in online Q&A communities, for example, demonstrate how ChatGPT is perceived when human collective intelligence is unleashed. We focus on textual data collected from the Stack Exchange community to investigate the major topics and sentiments generated when community members share their knowledge about ChatGPT. Different topic modeling and sentiment analysis algorithms were applied and compared to understand major perceptions and attitudes toward ChatGPT, and to illustrate the perceived opportunities and challenges surrounding ChatGPT and AI algorithms. Our study presents the differences in terms of ChatGPT-related sentiments and discussion themes across different types of sub-communities and highlights the critical roles of feature extraction and threshold selection in further improving the accuracy of topic models and precisely capturing the patterns of prevalent sentiments.

Ähnliche Arbeiten