Public perception and social acceptance of ChatGPT among Redditors: data-driven computationally intensive approach

Considering the increasing debates on ChatGPT in both technological and non-technological communities, this study aims to understand the public's perceptions of this AI-powered chatbot over time and the determinants of social acceptance of this technology, without which pertinent research and practice could be impeded. A data-driven computationally intensive approach was designed, based on which the ChatGPT-related posts and comments were collected from the Reddit social media platform at three different time points in recent months and years. Then, a topic modeling algorithm and two sentiment analysis techniques were employed to analyze the data. Finally, the topic modeling results were thematically analyzed. We discovered that although Reddit users' perceptions were slightly negative toward ChatGPT a short time after its introduction, the opinions are gradually becoming positive. Moreover, it was known that the public's conversations about this new technology fall into five main topics. Furthermore, a theoretical framework was inductively developed, comprising eight propositions on the determinants of social acceptance of ChatGPT. The identified determinants include perceived anthropomorphism, personalization, expertness, trustworthiness, usefulness, risks, transparency and enjoyment. The findings of this research offer insightful implications for ChatGPT users and developers as well as for AI researchers, practitioners and policymakers. For instance, the findings strengthen the conversation around investing in new interface designs that introduce more human-like, personalized and enjoyable elements. This is the first study to assess the temporal evolution of public perceptions regarding ChatGPT on the Reddit social media platform to the best of our knowledge.