Investigating the Multifaceted Consumer Perceptions on the Role of Artificial Intelligence (AI) in Social Media Interactions: A Topic Modelling-Based Exploration and Trust Perspective Sohaib Mohanna *, Ashraf Alhujailli *, Mohammad Aamir Khan *, Mohammed Ashar Khan *, Shafeeq Hussain Vazhathodi *, Abdullah Basiouni * Management Sciences Department, Yanbu Industrial College, Saudi Arabia * Abstract: This study investigates affective and cognitive dynamics within YouTube user interactions concerning artificial intelligence (AI) and correlates these dynamics with technology adoption frameworks. Commentary data extracted from YouTube videos on AI underwent analysis utilizing BERTopic, an advanced natural language processing model employing BERT embeddings. The model underwent fine-tuning through iterative optimization of parameters, including dimensionality reduction, clustering, and term weighting. Manual validation was conducted to affirm the machine generated topic coherence and relevance. The refined BERTopic model accurately pinpointed prominent themes in YouTube comments, emphasizing the pivotal roles of emotions and trust in user dialogues on AI. Qualitative manual analysis validated the model's discernment of subtle thematic nuances. The results underscore BERTopic's efficacy in scrutinizing vast quantities of user-generated content, unveiling intricate sentiment patterns. This offers valuable insights into the cognitive and affective dimensions influencing user interaction with nascent technologies such as AI. This research amalgamates cutting-edge natural language processing with human assessment to delve deeply into online user discourse surrounding AI. The adoption of a mixed-methods approach enhances comprehension of technology inclination, adoption trends, and online user involvement. Keywords: Artificial Intelligence, Emotions, Adoption, Trust Page: 1-20 Size: 863 KB Download PDF DOI: https://doi.org/10.51659/josi.24.221
