The Influence of AI-Generated Content on Public Perception and Trust
Students & Supervisors
Student Authors
Supervisors
Abstract
Background: In light of the rapidly evolving nature of generative artificial intelligence (AI), AI-generated content (such as deepfake videos, simulated images, and text generated by language models) has become more commonplace online. While such material is often indistinguishable from material produced by humans, concerns are emerging about its potential to inform, deceive and influence public opinion. Objective: It examines the effects of being exposed to artificially generated media on public perception and trust in digital contexts, notably in social media and online news and also the extent of individual ability to detect ‘AI’ generated media and the psychological and behavioral consequences of realizing that an information source trusted is artificial. Methodology: Both primary and secondary data were used in this study. Primary data were collected through surveys of internet users and in-depth interviews. Secondary data were gathered from academic articles, journals, and online sources related to AI-generated content and public perception. Key Findings: 64% of survey respondents could not differentiate AI-generated news from real sources. They believed digitally generated news to be trustworthy. Higher media exposure was associated with low critical engagement. In interviews, participants reported decreased trust in digital platforms once they learned they had been reading AI-generated content. The inclusion of fact checking labels and transparency in sources significantly improved user skepticism and engagement. Implications: Using four case studies, we show that artificial intelligence-generated media can prejudicially affect the perception of people in society, particularly within their lower digital literacy levels. We also note the urgent need for ethical policy, transparency measures (containers used to label and categorize artificial intelligence content), and public education to tackle such potential harms. Limitations: Most importantly the study was done in a specific population and geographical area which limits its generalizability. There is also the potential for bias as the data were self-reported. Future research will likely investigate long-term behavioral consequences and cross culture comparisons.
Keywords
Publication Details
- Type of Publication:
- Conference Name: The 4th International Conference on Information and Knowledge Management
- Date of Conference: 16/09/2025 - 16/09/2025
- Venue: East West University, Bangladesh
- Organizer: Department of Information Studies