This study introduces natural language explanations designed to promote more thoughtful reasoning about online videos and increase awareness of potentially biased or misleading content. We developed an end-to-end pipeline that extracts reflection triggers about video sources, topics, emotions, and sentiment to help users actively evaluate video usefulness. Our between-subjects study examining controversial topics reveals that users’ alignment with video messages significantly impacts perceived usefulness. While explanations were rated highly, they didn’t alter overall usefulness perceptions compared to videos alone. Notably, users with extreme negative alignment found videos less useful regardless of explanations and were more confident in their assessments. We interpret these findings through cognitive dissonance theory and propose design implications for explanations aimed at raising awareness about online deception.
Oana Inel,
Tomislav Đuričić,
Harmanpreet Kaur,
Elisabeth Lex,
Nava Tintarev
