AI Application Gives Users Real-Time Feedback on the Level of Peace in the Social Media Videos They Watch

The article presents a novel application that provides real-time feedback on the level of peace in social media videos, specifically YouTube content. It employs a combination of supervised machine learning, including a fully connected feedforward network and two convolutional neural networks, achieving approximately 97% accuracy in predicting peace levels from written text, while also integrating sentiment analysis and six context-level large language models (LLMs) to assess five social dimensions related to peace, achieving an \( r^2 \) correlation of approximately 0.60 with human coders. This development is significant for practitioners as it offers a tool for analyzing the impact of media on societal peace, enabling users and content creators to refine their media consumption and production strategies based on real-time feedback.