Media Effects Research Lab - Research Archive
Differences in Perception of Memes Based on Source: AI vs. Human
Student Researcher(s)
Clin KY Lai (Ph.D Candidate);
Zheng Yao (Ph.D Candidate);
Faculty Supervisor
INTRODUCTION
Memes are iconic cultural symbols with the potential to ‘go viral’ and hence enact impact on internet discourse and affect behaviors. With the rise of generative AI, it is increasingly difficult to distinguish between human-made and AI-generated content, which influences users' perceptions and trust in contents on the internet. Large social media companies such as Meta and Tiktok have proposed labeling contents as AI-generated as a solution to counter this. However, will such labels affect the perception of memes’ characteristics, especially those used to be considered as unique to humans?
RESEARCH QUESTION / HYPOTHESES
H1: Memes labeled as human-generated will be have higher perceived: a) creativity, b) humor, c) positive and negative emotions, d) virality, and e) sharing intentions as compared to those labeled as AI-generated
H2: Positive machine heuristics will moderate the relationship between the condition (human-as-source vs ai-as-source condition) and perceived: a) creativity, b) humor, c) positive and negative emotions, d) virality, and e) sharing intentions
H3: Negative machine heuristics will moderate the relationship between the condition (human-as-source vs ai-as-source condition) and perceived: a) creativity, b) humor, c) positive and negative emotions, d) virality, and e) sharing intentions
METHOD
Drawing upon the HAII-TIME theoretical framework, we explore how users perceive memes attributed to artificial intelligence (AI) or humans. In this experiment, before viewing the selected memes, participants were either told they will be viewing AI generated memes or human user generated memes. Then participants reported their perception of memes’ a) creativity, b) humor, c) positive and negative emotions, d) virality, e) their sharing intentions and f) machine heuristics. Multiple ANCOVA analyses were adopted to examine the main effect of source attribution and the moderation effects of machine heuristics.
RESULTS
Contrary to initial hypotheses, source labeling had no impact on participant evaluations, suggesting that content take precedence over creator attribution in meme perception. However, negative machine heuristics emerged as a potential moderator. Users with negative views of AI were less likely to perceive AI-generated memes as humorous, F(1, 114) = 3.69, p = .06 or creative, F(1, 114) =3.87, p = .05. Unexpectedly, positive machine heuristics also appear to play a significant role, predicting perceived virality, F(1, 114) = 4.60, p = .03, humor, F(1, 115) = 3.93, p =.05, and negative affect, F(1, 114) = 5.05, p = .03.
CONCLUSIONS/DISCUSSION
This research investigated the interaction between AI-generated memes, user perceptions, and perceptions of online content (memes). While the study did not conclusively support the hypothesis that source label (human vs. AI) significantly impacts meme evaluation, it suggests a robustness of meme perception, where content attributes outweigh creator attribution in driving perceptions. However, the research identified negative machine heuristics as a potential moderator. Users with negative views of AI were less likely to perceive AI-generated memes as funny or creative. Interestingly, positive machine heuristics also emerged as an unexpected predictor of perceived virality, humor, and even negative affect. From a practical standpoint, the findings highlight the importance of more research into how the labeling for AI-generated content affects user perceptions. Further research is warranted to explore the multifaceted effects of content labeling on user perceptions and behaviors across diverse online environments. By addressing these limitations and pursuing future research avenues, they offer valuable insights that can contribute to the ongoing conversation about how content labels shape user perceptions and behaviors.
For more details regarding the study contact
Dr. S. Shyam Sundar by e-mail at sss12@psu.edu or by telephone at (814) 865-2173