AI now creates more articles than humans, study finds

AI now writes more articles

An analysis by SEO firm Graphite found that AI-generated articles outnumbered human-written articles by volume in November 2024. The increase in AI-authored articles accelerated after ChatGPT went public in November 2022 and rose quickly through 2023.

This trend shows AI transforming content creation. Many publishers adopted AI to cut costs and speed up content production, sometimes replacing routine reporting and low-value pieces that previously went to paid writers.

Growth has plateaued recently, hinting that the era of rapid AI expansion may be leveling off.

Explosive growth since ChatGPT

Graphite’s data indicate the share of AI-authored articles rose from roughly 5% before ChatGPT to about 50% by late 2024.

Companies used AI to produce content for Google Search, social media, and ads, seeing it as a faster, cheaper alternative to traditional writing.

The speed of adoption surprised researchers. Within a year, AI accounted for almost half of all articles published online. Growth has slowed, suggesting companies are weighing AI efficiency against performance in search engines and other platforms.

AI content plateauing now

While AI content exploded initially, the percentage of new AI articles has stabilized since May 2024. AI and human-written articles are roughly split fifty-fifty. Plateauing may be linked to AI content farms seeing many articles fail to rank in Google or appear in ChatGPT results.

This suggests human-written content is still critical for visibility. Search engines favor quality human writing, making AI alone insufficient for dominating web search traffic despite high volume.

Detecting AI articles is tricky

Graphite used Surfer to classify articles as AI-generated if half or more of the content was written by a large language model. Graphite estimated a false positive rate of about 4.2% and a false negative rate of about 0.6%.

Even with these tools, perfect detection is impossible. Many writers blend AI with human editing, which can fool detection software. It’s hard to know how much content is fully AI-created versus AI-assisted.

Quality keeps improving fast

Studies show AI content is becoming harder to distinguish from human writing. In many cases, AI-generated articles are just as good or better than human-written ones. This makes it easier for companies to adopt AI at scale.

As AI models improve, adoption may rise further. Quality improvements blur lines between human and AI work, raising questions about originality and reliability across the web.

Human visibility still matters

Despite AI prevalence, most articles seen in Google search are human-written. A report found 86% of articles ranking in Google were by humans, 14% AI-generated. Visibility isn’t determined by volume alone.

This gap may push AI publishers to rethink strategies. Producing thousands of AI articles may not guarantee audience reach, emphasizing the continued value of human editorial judgment.

AI-assisted human editing

Many creators combine AI drafts with human editing. They let AI generate a first version, then refine it manually. This can bypass detection tools and improve quality, producing content that’s partially AI but polished.

This hybrid model may become standard. The line between AI-generated and human-edited content continues to blur, challenging the classification and evaluation of web content.

Common Crawl dataset used

Graphite sampled 65,000 English language URLs from Common Crawl between January 2020 and May 2025, restricting the sample to articles that were more than 100 words and that included article schema markup.

This ensures a broad view of web trends. Some paywalled or restricted sites weren’t included, so human-written content could be underrepresented.

LLMs power content growth

Large language models like ChatGPT, Claude, and Gemini drove the AI article surge. They generate content quickly on diverse topics, making it easier for companies to publish more articles than ever.

LLMs allow highly scalable content creation. Companies can target multiple channels simultaneously, producing SEO-optimized and social-friendly articles, changing the economics of publishing.

Plateau hints at limits

Despite AI growth, its share of new articles has plateaued since late 2024. Search engines prefer human content, and mass AI publishing has a limited impact on reader engagement.

The plateau signals AI cannot fully dominate content ecosystems. Companies may need human oversight to maintain visibility, not just churn AI articles.

Training data challenges

AI models rely on massive datasets like Common Crawl. Many sites now block crawlers, especially paywalled content, excluding high-quality human articles from AI training.

This may explain why some AI articles underperform. Without access to all high-quality human content, AI models miss nuances influencing ranking and engagement.

Hybrid strategies rising

Hybrid workflows combining AI drafts with human editing are gaining traction. They scale production without losing quality and make finished content appear fully human-authored.

This may represent the future of publishing, balancing AI efficiency with human judgment. AI detection tools may underestimate AI involvement in hybrid articles.

The evolving landscape

The web is a mix of AI and human content. As AI improves, distinctions blur. Readers may encounter AI-influenced content without realizing it, changing expectations for quality online.

This evolving landscape challenges publishers, search engines, and readers. Keeping track of AI’s impact is crucial for navigating content creation and consumption.

Could AI put lives at risk sooner than we think? Don’t miss how the ChatGPT lawsuit over teen suicide could spark a big tech reckoning.

Humans and AI coexist online

AI dominates volume, but human content remains critical for search visibility and trust. Hybrid strategies combining AI and human editing are becoming common, creating a symbiotic approach.

AI and human content will continue to coexist. Focus will shift from replacement to collaboration, ensuring content is fast and valuable for readers.

What if your next work meeting felt more like stepping into a shared virtual space? Don’t miss how Meta is using AI to make VR meetings feel closer to reality.

Surprised that machines are leading in article writing? Hit like and tell us your take in the comments.

Read More From This Brand:

• OpenAI’s AI browser rollout is expected soon
• OpenAI engineer quits, describing the company as complete chaos
• DeepSeek says R1 now rivals OpenAI and Gemini models

Don’t forget to follow us for more exclusive content right here on MSN.

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one.

This slideshow was made with AI assistance and human editing.