<![CDATA[The development of artificial intelligence (AI) has generated extensive discussions across various social media platforms. Therefore, understanding public sentiment toward AI is crucial for assessing public perception of the new technology. This study analyzes AI-related sentiment across multiple platforms, including X, Reddit, TikTok, and YouTube. A large-scale dataset consisting of over 70,000 comments was collected and analyzed using both transformer-based and traditional machine learning approaches. This study uses XLM-RoBERTa as the primary model, while IndoBERT and SVM with TF-IDF served as baselines for comparison. Results show that IndoBERT achieves the highest accuracy on Indonesian-language data, highlighting the effectiveness of language-specific models. Meanwhile, XLM-RoBERTa demonstrates strong performance across multilingual datasets, demonstrating its robustness in handling heterogeneous social media content. In contrast, the baseline SVM with TF-IDF exhibits lower performance due to limited contextual understanding. These results demonstrate the superiority of transformer-based models for multiplatform sentiment analysis of discussions about artificial intelligence.]]>
Copyrights © 2026
