Anthology: AI-Augmented Sensemaking Interface to Reduce Radicalization in Conversations
Om Gokhale, Chayapatr Archiwaranguprok, Rachel Poonsiriwong, Pattie Maes, and Pat Pataranutaporn. 2026. Anthology. In Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA ’26). Association for Computing Machinery, New York, NY, USA, Article 91, 1–6. https://doi.org/10.1145/3772363.3799062
Abstract
As radicalization intensifies, people increasingly encounter multi-perspective conversations through recordings, transcripts, and online forums, yet these formats offer no structural support for synthesis, requiring users to mentally reconstruct thematic patterns while often stripping away the human voices that enable empathy. We present Anthology, a prototype exploring how AI-augmented semantic organization transforms recorded conversations into interactive, semantically-clustered visualizations that invite users to hear unfamiliar perspectives in speakers’ own voices. Through a between-subjects experiment (N=60), we demonstrate that Anthology significantly reduces position extremity (p=.022, d=0.60), defined as how far someone’s stance on an issue is from the middle, compared to traditional transcripts. We discuss implications for designing conversation interface that foster critical engagement with diverse viewpoints in contexts such as online discussion space, citizens’ assemblies, and deliberative forums.