Paper Type

ERF

Abstract

Marketers are increasingly adding “AI origin” labels to ads, but those small tags can change what people look at, how they feel, and what they take away. This study examines three practical levers: whether a label is shown, how it is worded (“human made”, “AI assisted”, “AI generated”), and when it appears (pre , concurrent, post exposure). In three within subject lab experiments, participants view image and video ads while we record eye movements (time to first fixation; dwell on label, brand, and product areas), facial expressions, and galvanic skin response, followed by measures of ad/brand attitude and perceived authenticity. Labels capture early fixations and pull attention away from persuasive content; stronger automation wording carries steeper authenticity and attitude costs; and pre‑exposure timing intensifies these effects. By tying real time biometric signals to downstream judgments, the work offers practical guidance on how to disclose AI use in ways that protect transparency without needlessly hurting performance.

Paper Number

1631

Comments

SIG HCI

Download

Share

COinS
 
Aug 15th, 12:00 AM

The Cost of Transparency: How AI Disclosure Shapes Consumer Responses in Digital Advertising

Marketers are increasingly adding “AI origin” labels to ads, but those small tags can change what people look at, how they feel, and what they take away. This study examines three practical levers: whether a label is shown, how it is worded (“human made”, “AI assisted”, “AI generated”), and when it appears (pre , concurrent, post exposure). In three within subject lab experiments, participants view image and video ads while we record eye movements (time to first fixation; dwell on label, brand, and product areas), facial expressions, and galvanic skin response, followed by measures of ad/brand attitude and perceived authenticity. Labels capture early fixations and pull attention away from persuasive content; stronger automation wording carries steeper authenticity and attitude costs; and pre‑exposure timing intensifies these effects. By tying real time biometric signals to downstream judgments, the work offers practical guidance on how to disclose AI use in ways that protect transparency without needlessly hurting performance.