Appeal and Scope of Misinformation Spread by AI Agents and Humans

2025-05-07

AI Summary

Existing studies lack operational metrics to quantify the differential impact of human- versus AI-generated COVID-19 vaccine misinformation on social media. Method: We propose a dual-dimensional framework—“Appeal” (measured via user engagement and network centrality) and “Scope” (measured via structural reach)—applied to 5.8 million vaccine-related tweets. Using Tweedie regression, centrality analysis, and joint temporal–source modeling, we systematically compare dissemination dynamics. Contribution/Results: Our analysis reveals distinct phase-dependent patterns: human-generated misinformation peaks in harmfulness during the first week post-vaccine rollout, whereas AI agents exert strongest influence during early R&D phases; misinformation prevalence is significantly higher in the first two phases than in later stages. This interpretable, reusable framework advances evidence-based misinformation governance by enabling granular, source-aware impact assessment.

Technology Category

• Application Domains: Misinformation & Fake News
• Humans and AI: Planning and Decision Support for Human-Machine Teams
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Responsible Web: Mitigating misinformation and disinformation
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

This work examines the influence of misinformation and the role of AI agents, called bots, on social network platforms. To quantify the impact of misinformation, it proposes two new metrics based on attributes of tweet engagement and user network position: Appeal, which measures the popularity of the tweet, and Scope, which measures the potential reach of the tweet. In addition, it analyzes 5.8 million misinformation tweets on the COVID-19 vaccine discourse over three time periods: Pre-Vaccine, Vaccine Launch, and Post-Vaccine. Results show that misinformation was more prevalent during the first two periods. Human-generated misinformation tweets tend to have higher appeal and scope compared to bot-generated ones. Tweedie regression analysis reveals that human-generated misinformation tweets were most concerning during Vaccine Launch week, whereas bot-generated misinformation reached its highest appeal and scope during the Pre-Vaccine period.

Problem

• Examining AI and human misinformation impact on social networks
• Proposing metrics to measure tweet appeal and reach
• Analyzing COVID-19 vaccine misinformation trends over time

Innovation

• Proposes Appeal and Scope metrics for misinformation
• Analyzes 5.8 million COVID-19 vaccine tweets
• Uses Tweedie regression to compare human and bot tweets