Evaluating AI-Generated Satire against Human-Written Content: A Comparative Analysis

Abstract

This study examines the comparative quality of AI-generated satirical content against human-written satirical articles. Using a database of 160 satirical articles from Times New Roman as a baseline, we developed 20 AI-generated stimuli using Artificial Intelligence (AI) and ww evaluated the stimuli in comparison with real satire news using four large language models (LLM) (Deepseek, Claude, Gemini, and ChatGPT). Through empirical evaluation, we assessed whether AI-generated satirical content is perceived as more humorous than authentic human-written content. This research contributes to understanding the evolving capabilities of AI in creative content generation and its implications for humor perception, media production, and computational linguistics.