TY - GEN
T1 - AI-Generated or AI-Modified? User Reactions to Labeling AI Use in Social Media Posts
AU - Jung, Yongnam
AU - Hua, Peixin
AU - Bao, Jiaqi
AU - Sundar, S. Shyam
N1 - Publisher Copyright:
© 2025 Copyright held by the owner/author(s).
PY - 2025/4/26
Y1 - 2025/4/26
N2 - Although social media platforms are beginning to implement policies for labeling AI usage in posts, we do not know how this labeling affects user perceptions and which types of labeling source and language would be most effective. What does AI-generated content mean to users? What are users’ perceptions of the content, content creator, and platform that has AI-labeled content? Do the effects differ depending on whether the label is attributed to the user or the platform? A focus group study (N=14) revealed that users appreciate how AI helps to create better content. However, their perceptions of AI-labeled content are shaped by their mental models of how social media algorithms work. Some participants viewed AI-labeled content as more trendy, while others saw it as direct advertisements. Some believed the content to be automatically fake, but their reactions varied depending on the type of content or account. Regarding labeling source, users preferred self-labeling over platform labeling. We discuss theoretical and practical implications for the design of social media interfaces for disclosing AI usage.
AB - Although social media platforms are beginning to implement policies for labeling AI usage in posts, we do not know how this labeling affects user perceptions and which types of labeling source and language would be most effective. What does AI-generated content mean to users? What are users’ perceptions of the content, content creator, and platform that has AI-labeled content? Do the effects differ depending on whether the label is attributed to the user or the platform? A focus group study (N=14) revealed that users appreciate how AI helps to create better content. However, their perceptions of AI-labeled content are shaped by their mental models of how social media algorithms work. Some participants viewed AI-labeled content as more trendy, while others saw it as direct advertisements. Some believed the content to be automatically fake, but their reactions varied depending on the type of content or account. Regarding labeling source, users preferred self-labeling over platform labeling. We discuss theoretical and practical implications for the design of social media interfaces for disclosing AI usage.
UR - https://www.scopus.com/pages/publications/105005755969
UR - https://www.scopus.com/inward/citedby.url?scp=105005755969&partnerID=8YFLogxK
U2 - 10.1145/3706599.3720264
DO - 10.1145/3706599.3720264
M3 - Conference contribution
AN - SCOPUS:105005755969
T3 - Conference on Human Factors in Computing Systems - Proceedings
BT - CHI EA 2025 - Extended Abstracts of the 2025 CHI Conference on Human Factors in Computing Systems
PB - Association for Computing Machinery
T2 - 2025 CHI Conference on Human Factors in Computing Systems, CHI EA 2025
Y2 - 26 April 2025 through 1 May 2025
ER -