Estimating the Percentage Growth of AI Content Across Categories of the Web

Abstract

The rapid adoption of generative AI has made distinguishing authentic from artificially generated web content increasingly difficult, yet little research has tracked how this proportion changes over time. Prior works have focused primarily on detecting AI-generated text, with no study systematically measuring and forecasting growth across multiple web content categories over time. This research quantifies AI-generated web content, specifically for blogs, social media, and news, using a two-phase approach: ML classification (traditional vs. deep learning) for labeling, and time-series forecasting (linear vs. non-linear) for tracking growth. Across the evaluated sources, news and social media averaged around 60% in AI-generated content or heavy automation use, while blogs averaged in the range of 20-30% through the classifier trained in the research. Finally, based on the result of the best linear model displaying the predicted line, AI content has been shown to grow a slight amount, with blogs increasing 0.14%, 0.078% for news, and 0.032% for social media every month since 2020. These findings suggest that content consumers, platform moderators, and policymakers should exercise more caution, especially as AI-generated content, particularly in news will continue to grow in the future and may amplify misinformation at scale.