Sounding Trustworthy: AI-Generated Audio Outperforms Video and Images in Political Communication
Abstract
Artificial Intelligence (AI) is transforming political communication through AI-generated content, including deepfake videos, synthetic voices, and digitally manipulated images. While these advancements offer new opportunities for engagement, they also raise concerns about misinformation and political trust. This study investigates the effects of AI-generated media formats on individuals' willingness to follow political recommendations and the role of media realism in shaping trust. Through an online experiment, 150 participants assessed political content in varying degrees of realism across audio, video, and image formats. Results were analyzed using a combination of linear mixed effects analysis and natural language processing, and indicate that AI-generated audio is perceived as more trustworthy than image or video content, while lower realism levels trigger skepticism. These findings contribute to discussions on political AI, emphasizing the need for media literacy and regulatory frameworks to mitigate misinformation risks.
Study specs
An online experiment with participants assessing AI-generated political content in audio, video, and image formats; data analyzed using linear mixed effects analysis and NLP.
- Institution
- Dalhousie University
- Discipline
- Artificial Intelligence,Cognitive Science
- Sample Size
- N=150
- Study Type
- Experimental Study
- Year
- 2025
- Human Data Platform
- Prolific
- Source
- View Source Google Scholar
Measured Outcomes
Impact of AI-generated media formats on trust and willingness to follow political recommendations, considering realism levels.
Peer Review & Critical Discussion
Potential Selection Bias in 2023 Cohort
The participant pool shows a concerning overrepresentation of users from high-income demographics. Looking at Table 3, we can see that 78% of respondents had annual incomes above $75k, which significantly limits the generalizability of these findings to broader populations.
Non-naive Participants Issue
I've noticed a methodological concern regarding participant naivety. Given that Prolific users often complete multiple studies, there's a real risk that participants had prior exposure to similar experimental paradigms, which could confound the results.
RLHF Applicability to This Study Design
The implications for RLHF training pipelines are understated. If we accept the authors' conclusions about preference stability, this has direct consequences for how we should structure reward model training. The temporal decay effect described in Section 4.2 is particularly relevant.
Verify your expertise to join discussion
Create an account and verify your credentials to participate in peer discussions.