Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Examining the Impact of Label Detail and Content Stakes on User Perceptions of AI-Generated Images on Social Media

Jingruo Chen, TungYen Wang, Marie Williams, Natalia Jordan, Mingyi Shao, Linda Zhang, Susan R. Fussell

TL;DR

This study addresses trust and authenticity concerns surrounding AI-generated images on social media by examining how label detail and content stakes influence user perceptions and engagement. Using a two-factor, within-subjects design with three label detail levels and two stake conditions, the authors assess engagement, believability, manipulation, and label helpfulness among 105 participants. Results show that more detailed labels increase perceived transparency and trust without reducing engagement, while low-stakes content evolves toward higher engagement and believability and high-stakes content prompts more information seeking and skepticism. The findings imply that platforms can adopt detailed labeling strategies to enhance transparency without sacrificing engagement, offering practical guidance for labeling AI-generated media in social feeds.

Abstract

AI-generated images are increasingly prevalent on social media, raising concerns about trust and authenticity. This study investigates how different levels of label detail (basic, moderate, maximum) and content stakes (high vs. low) influence user engagement with and perceptions of AI-generated images through a within-subjects experimental study with 105 participants. Our findings reveal that increasing label detail enhances user perceptions of label transparency but does not affect user engagement. However, content stakes significantly impact user engagement and perceptions, with users demonstrating higher engagement and trust in low-stakes images. These results suggest that social media platforms can adopt detailed labels to improve transparency without compromising user engagement, offering insights for effective labeling strategies for AI-generated content.

Examining the Impact of Label Detail and Content Stakes on User Perceptions of AI-Generated Images on Social Media

TL;DR

This study addresses trust and authenticity concerns surrounding AI-generated images on social media by examining how label detail and content stakes influence user perceptions and engagement. Using a two-factor, within-subjects design with three label detail levels and two stake conditions, the authors assess engagement, believability, manipulation, and label helpfulness among 105 participants. Results show that more detailed labels increase perceived transparency and trust without reducing engagement, while low-stakes content evolves toward higher engagement and believability and high-stakes content prompts more information seeking and skepticism. The findings imply that platforms can adopt detailed labeling strategies to enhance transparency without sacrificing engagement, offering practical guidance for labeling AI-generated media in social feeds.

Abstract

AI-generated images are increasingly prevalent on social media, raising concerns about trust and authenticity. This study investigates how different levels of label detail (basic, moderate, maximum) and content stakes (high vs. low) influence user engagement with and perceptions of AI-generated images through a within-subjects experimental study with 105 participants. Our findings reveal that increasing label detail enhances user perceptions of label transparency but does not affect user engagement. However, content stakes significantly impact user engagement and perceptions, with users demonstrating higher engagement and trust in low-stakes images. These results suggest that social media platforms can adopt detailed labels to improve transparency without compromising user engagement, offering insights for effective labeling strategies for AI-generated content.
Paper Structure (17 sections, 2 figures, 1 table)

This paper contains 17 sections, 2 figures, 1 table.

Table of Contents

  1. Background
  2. Method
  3. Participants
  4. Materials
  5. Image Selection
  6. Label Design
  7. Procedure
  8. Measures
  9. Data Analysis
  10. Findings
  11. RQ1: Impact of Different Levels of Label Detail
  12. RQ2: Impact of Different Contents (High stake v.s. Low stake) of AI-generated Images
  13. RQ3: User Preferences regarding labeling for AI-generated Images
  14. Discussion
  15. Design Implications
  16. ...and 2 more sections

Figures (2)

  • Figure 1: Three AI image label levels based on the SAT transparency model: Basic describes AI tool use (Level 1: Goals & Actions), Moderate adds content reasoning and model used (Level 2: Reasoning), and Maximum includes broader implications (Level 3: Projections).
  • Figure 2: AI-generated and real image stimuli by content stakes. AI-generated images included low-stakes (pets, mountains, portraits) and high-stakes (protests, law enforcement, futuristic scenes); corresponding real-image fillers reduced priming and preserved ecological validity.