The AI writing revolution has seemingly hit a plateau, according to a recent analysis by Graphite, a digital marketing agency. The study reveals that the percentage of online news articles, blog posts, and listicles primarily generated by AI has stabilized at around 50% for over a year. This finding challenges the notion that AI would completely replace human writers online, at least for now.
The initial surge in AI-generated content can be attributed to the launch of ChatGPT in November 2022. Within a year, AI-written articles accounted for 35.9% of new online articles, and this number rose to 48% within two years. However, since early 2025, the growth has stagnated, indicating a potential limit to AI's dominance in content creation.
Dan Klein, a UC Berkeley professor and AI model CTO, highlights a critical concern: the internet's reliance on AI-generated content. He warns that if we continue to create knowledge that is dependent on these models, the web could become a feedback loop of low-quality, machine-generated content. This raises questions about the sustainability and authenticity of AI-written material.
Graphite's methodology involved sampling 55,400 English-language URLs from Common Crawl, focusing on articles and listicles published between January 2020 and March 2026. The analysis used AI-checking tools to determine the primary source of each article's content. However, the line between human and AI writing remains blurred, as many articles are a blend of both.
The quality of AI-generated content is improving, and it often rivals human-written pieces. This makes it challenging for the general public to discern whether an article is AI-created or not. As AI continues to evolve, its impact on the online writing landscape will be a topic of ongoing interest and debate.
In conclusion, while AI now contributes significantly to online content creation, the plateau in its growth suggests a temporary equilibrium. The future of AI in writing remains uncertain, and further research is needed to understand the long-term implications for human writers and the quality of online content.
