The PAGE framework turns artificial intelligence into an instrument for comics research. It first parses each scanned page into panels, text, and a bibliographic record; then annotates every page with a vision‑language model that follows a written codebook, mimicking a human coder; next it grounds a subset of these annotations in human‑generated labels for calibration; finally it explains changes across an archive while controlling for publisher and genre.
The framework was demonstrated on 3,929 American comic books (1938‑1956) and 87 volumes of Japanese manga, yielding three separate studies. The first pairs text search with visual reading to uncover how a mass medium portrayed political power. The second quantifies the visual form of pages across an industry and recovers the stylistic hand of individual artists. The third tracks who speaks in the two comics traditions and validates the results against human annotation.
Beyond scholarship, PAGE can make archives searchable by drawn content, open comics to blind and low‑vision readers, aid preservation of fragile collections, and credit uncredited artists. The authors outline a research agenda in which AI further strengthens the study, preservation, and reach of the medium.
Review: PAGE showcases AI’s potential for large‑scale, cross‑cultural heritage analysis, providing a systematic, reproducible pipeline for comics studies while opening new avenues for public access and cultural preservation.