AI Displaces Human Content, Shifts Cultural Power
GPT-3's 19.1% Persuasion Increase
A Stanford University HAI Policy Brief reported that GPT-3-generated articles were highly persuasive, increasing participant agreement by 19.1 percentage points from a 24.4% baseline. Empirical evidence from the 2024 elections, however, suggested that traditional sources of influence often overshadowed AI's impact, according to the Knight First Amendment Institute at Columbia University. GenAI applications provide highly personalized interactions that amplify confirmation bias by reinforcing pre-existing beliefs and obscuring consensus, as detailed in the Annals of the New York Academy of Sciences. Through sycophantic responses, AI validates user biases, allowing users to steer narratives toward preferred outcomes with minimal effort, the Annals of the York Academy of Sciences found. A preprint study indicated that AI systems can adjust manipulation strategies in real time based on user input, enabling more targeted persuasion over prolonged interactions. The Stanford University HAI Policy Brief concluded that while human-written propaganda increased agreement by 23 percentage points, AI content, when combined with human-machine teaming, proved just as persuasive—or even more persuasive—than real-world propaganda.
Nasser Bahrami's 'AIgemony'
This creates unobservable power hierarchies through "AIgemony," which Nasser Bahrami describes as "a form of dominance embedded in intangible data that is harder to understand or counteract than classical hegemony." The shift toward AI-driven bias formation necessitates a move from top-down institutional oversight to decentralized algorithmic governance, driven by the inadequacy of traditional market-based regulatory frameworks, Policy Review argues. This transition involves a new model of public-private, multi-stakeholder governance, as highlighted by the Center for Democracy & Technology and Johns Hopkins SAIS. Algorithmic governance of data flows allows platforms to trade free services for user-generated data, establishing structural control over information ecosystems, Policy Review explained. Nasser Bahrami observed that dominant narratives shift from static models to recursive, AI-driven systems that continuously reshape based on real-time data. These mechanisms compromise auditability due to limited observational data from private GenAI interactions, a lack of human review for vast AI-user interactions, and the difficulty of recognizing AI-generated content, according to the Annals of the New York Academy of Sciences, a preprint study, and WJARR.
AI's Credibility Deficit
Frontiers in Psychology determined that content perceived as machine-generated elicits lower trust, increased error sensitivity, and more conservative quality ratings. This "credibility deficit" can limit its persuasive efficacy if its artificial origin is identified, Frontiers in Psychology found. Despite this, WJARR documented that AI selectively drains institutional authority from traditional archives and conventional media by eroding public trust in authority organizations. Simultaneously, the Annals of the New York Academy of Sciences and Newswise indicate that synthetic media artificially inflates the perceived legitimacy of algorithmic narratives by mass-producing content that reinforces echo chambers and manufactures "synthetic consensus."
AIgemony Reshapes Information Governance
WJARR and Nasser Bahrami assert that generative AI's rapid hypercustomization is fundamentally reorienting information ecosystem governance and shifting structural power dynamics, eroding public trust in conventional media. This rebalancing is crucial to counteract "AIgemony," which embeds dominance in intangible data and reshapes information environments. Policy Review and Nasser Bahrami argue that the move from static, analog narratives to recursive, AI-driven dominance necessitates new governance models that treat platform power as societal infrastructure.
Comments ()