Visual Information in the Age of AI: Evidence from Corporate Executive Presentations

Abstract

This paper constructs and studies a comprehensive data set comprised of corporate executive presentations. Executive presentations are unique in that they provide an abundance of visual information about a firm’s project designs and production plans. In the aggregate, these presentations allow us to explore the value of visual information and examine how market participants with varying levels of technological access respond to such information. Using a state-of-the-art deep learning model, we extract forward-looking operational information from presentation slide images. We find that short-term abnormal returns are positively associated with forward-looking operational information, but not with backward-looking or financial information. AI-equipped financial institutions respond strongly to visual signals, whereas other institutions and retail investors do not. Our study provides novel evidence that AI adoption rewards investors with an informational advantage, creating a potential AI divide among market participants.