Princeton University Users: If you would like to view a senior thesis while you are away from campus, you will need to connect to the campus network remotely via the Global Protect virtual private network (VPN). If you are not part of the University requesting a copy of a thesis, please note, all requests are processed manually by staff and will require additional time to process.

Browsing by Author "Lewis, Tierra E."

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Thumbnail Image

    The Quality Question: How AI Image Quality Impacts Perceived Motivations and Engagement Intentions on Social Media Platforms

    (2025) Lewis, Tierra E.; Guess, Andy

    Previous research and policy debates regarding the topic of static images generated by artificial intelligence platforms have explored the effects of photorealistic AI-generated media on public perceptions of the truth and the potential of these images to escalate online disinformation, particularly during elections in the United States. While this research is crucial to understanding the consequences of convincing AI-generated images and media online, the current quality of AI-generated images that everyday social media users create or interact with is often far from realistic. These low-quality images lack the critical aesthetic details often found in higher-quality AI-generated images. Yet, low-quality AI-generated images frequently receive high levels of engagement through likes, comments, or shares on social media platforms, such as Meta’s Facebook and Instagram or the platform X (formerly known as Twitter). Considering the contentious nature of creating low-quality AI-generated images in spite of increasing integration of generative artificial intelligence into online platforms, this type of media exists as a complex yet under-researched form of digital creation. This study collects survey data to address this gap in current research on artificial intelligence, emphasizing the role of low-quality AI-generated images. The objective is to assess how users perceive the motivations behind the creation of these images as well as their engagement patterns with these images based on the following attributes: believability, accuracy, informativeness, humor, and deception. Findings generally support increased negative perceptions towards low-quality, AI-generated images while high- quality, AI-generated images are generally found to be perceived as more believable, accurate, and informative. Advanced age and partisan affiliation also have significant effects on perceptions and engagement intentions with AI-generated media. Informed by these results, social media platforms should support the use of more labels across all AI-generated media. Furthermore, policymakers should pursue legislation that can enforce this method of moderation, protecting the rights of users, platforms, and creators in the process.