Is An Article Read Here Written By An AI Or A Human?

Depending on the source, artificial intelligence (AI) is increasingly being used to produce online content. Before proceeding further, a full disclosure. This blog site, since the first release of ChatGPT 3.0, has been exploring how generalized AI (GenAI) can assist in web content creation. Recently, the AI of choice has been Perplexity.ai, which is used by me to formulate questions and organize background research and source identification for articles like this one.

A May 19, 2025, survey that looked at 900,000 pages of web content published its results. The survey was done by Ahrefs, an SEO company. It shows that I am not alone in using AI. In fact, 74.2% of the pages analyzed contained AI-generated material.

Ahrefs built a content detector, a machine learning algorithm called “bot_or_not,” which compiled interesting results. The company acknowledged that the tool might not be 100% accurate, but it believes the data isn’t far from the truth.

Nevertheless, the findings make for interesting reading. Are they an accurate reflection of the entire web? Probably not considering all global content producers, but for North America, they are likely a good measuring stick.

What did bot_or_not discover?

• 2.5% of web pages were pure AI-produced content.
• 25.8% were pure human-produced.
• 71.7% were a mix of the two.

Where mixed content was found:

• 25.86% showed 11 to 14% AI use.
• 20.5% showed 41 to 70% AI use.
• 15.51% showed 71 to 99% AI use.

When 879 content marketers were asked if they used AI to help create online material, 87.37% said yes, while the balance reported not using AI in any capacity. Where was AI being used to produce this content? The graph below tells a story.

Geography plays a role in the percentage of AI-generated or AI-assisted content appearing on the web. The Anthropic Economic Index (Anthropic is a major AI player) has produced some interesting statistics on this subject, looking at large and small economies and countries with and without technologically advanced infrastructure.

The results show that high-income countries are much greater users of AI than emerging economies. In the U.S., although California produces the greatest volume of AI content, per capita usage of AI is highest in Washington, DC and Utah.

Low-income countries use AI disproportionately more to help with the writing of software code, with adoption rates for AI usage in business at 59% in India, 58% in the UAE, 53% in Singapore, and far less in Canada (37%), Japan (34%), U.S. (33%), Germany (32%) and Australia (29%). Workers in Latin America and Asia-Pacific countries show a greater openness to AI adoption at 48% and 46% respectively, higher than the numbers for North America at 36%.

A Three-Year AI Perspective, Past and Future

Today, AI isn’t just helping marketers and bloggers write copy. In the three years since ChatGPT 3.0 was released, GenAI, machine learning, and natural language models have become essential tools for business, medicine, science and technology.

Today’s AI is:

• Streamlining product development,
• Helping to sort out the kinks in supply chains,
• Creating a transportation revolution through self-driving, autonomous and connected vehicles,
• Transforming medical diagnostics and patient data analytics,
• Revolutionizing the development of new drugs,
• Accelerating all kinds of scientific discovery,
• Impacting conservation efforts for species preservation,
• Making fusion energy that much closer to becoming a commercial reality,
• And creating robots that are more “human.”

Looking three years into the future, tomorrow’s AI will:

• enable and accelerate scientific discovery,
• help in the fight to mitigate and adapt to climate change,
• enhance precision medicine and match diseases in people to personal cures,
• establish new norms and institutions for government, including decision and negotiation support, translation and other skills to remove barriers to resolving intractable disputes,
• And change institutional learning and education to meet individual lifelong needs.

The Downside of Our AI Future

If AI writes what you read, will AI take over thinking for us as well? The dark side of AI is already impacting employment. AI is making it possible for employers to use fewer workers, particularly in white-collar entry-level jobs like those of interpreters, translators, writers, customer service representatives, salespeople, reporters, telemarketers, technical writers, editors, bookkeepers, accountants, law clerks, web developers, librarians, and market researchers, who are at risk.

On the blue-collar side, the integration of AI into robotics will undoubtedly impact assembly line, logistics and warehousing workers. The hands-on trades like plumbing, HVAC and electricians will survive the AI and robotics revolutions. Construction workers, however, may see their numbers shrink as 3D-printed buildings and robotic drywallers and painters change the industry.

A final note, I used AI to assist me in sourcing the links in this posting. I would estimate that the ratio of human and AI content in this piece is 80/20 in my favour.