Does AI See like Art Historians? Interpreting How Vision Language Models Recognize Artistic Style

Marvin Limpijankit, Milad Alshomary, Yassin Oulad Daoud, Amith Ananthram, Tim Trombley, Elias Stengel-Eskin, Mohit Bansal, Noam M. Elcott, Kathleen McKeown · 2026

Read Does AI See like Art Historians? Interpreting How Vision Language Models Recognize Artistic Style by Marvin Limpijankit, Milad Alshomary, Yassin Oulad Daoud, Amith Ananthram, Tim Trombley, Elias Stengel-Eskin, Mohit Bansal, Noam M. Elcott, Kathleen McKeown (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-11024.txt.

Description

Marvin Limpijankit and coauthors ask whether vision-language models recognize artistic style as human experts do. The team pairs computer scientists with art historians. A latent-space decomposition pulls out the concepts driving style prediction. Art historians then judge them for coherence and relevance. Quantitative and causal checks back those judgments. Most extracted concepts prove coherent and meaningful. Most concepts behind a given prediction are judged relevant. Even stray successes often rest on formal features. One such feature is dark-light contrast.

Catalog record

CREATORMarvin Limpijankit, Milad Alshomary, Yassin Oulad Daoud, Amith Ananthram, Tim Trombley, Elias Stengel-Eskin, Mohit Bansal, Noam M. Elcott, Kathleen McKeown
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36535
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.11024

Browse the full library · About the Tower of Babel