Over a Third of New Web Content is AI. Here’s How to Stay Afloat.

The short version

A new study found that over a third of new web pages show signs of being written by AI. This explosion of low-effort, automated content—often called “AI slop”—is making it harder to find trustworthy information online. My strategy is to get better at spotting the fakes and double down on trusted, human-curated sources.

It’s a feeling I’ve had for months now. Search for a product review, a recipe, or an answer to a specific question, and the top results feel… off. The language is bland and repetitive, the advice is painfully generic, and there’s no trace of a real human personality. The articles are technically readable, but they’re completely soulless. They offer no real insight, just a collection of keywords stitched together to please an algorithm.

I knew I wasn’t imagining it, but I didn’t realize how big the problem was until now. The feeling of a web getting clogged with junk just got a hard number attached to it, and it’s a big one. This isn’t just about a few spammy sites anymore; it’s a fundamental shift in the information landscape we all navigate every day.

What did the Pew study actually find?

The core finding is startling: 35% of all new English-language web pages published after November 2022 show strong signs of being generated or heavily edited by artificial intelligence. The new research from the Pew Research Center analyzed a massive sample of 490,000 web pages from a third-party data provider to reach this conclusion. They looked for specific artifacts left behind by AI writing tools, creating a reliable method for flagging automated content.

This isn’t just happening on small, obscure blogs either. The study found that .com domains, the backbone of the commercial web, are a major source of this content. Think about that—for every three new articles, reviews, or guides you come across, one of them was likely written by a machine with minimal human oversight. This rapid increase confirms that we’ve entered an era of mass-produced, automated information.

What’s particularly telling is the timing. The study’s cutoff of November 2022 is, of course, the month ChatGPT was released to the public. The data shows a clear inflection point, where the cost and effort required to produce web content plummeted to near zero, and content farms immediately seized the opportunity. The result is the flood of mediocre content we’re all wading through right now.

Why is this a problem?

The issue isn’t AI itself. Artificial intelligence is a powerful tool. The problem is the economic incentive driving the creation of this content. This flood of AI-generated text is a direct consequence of an ad-based internet model where clicks equal revenue, regardless of the quality of the information that earned the click. It’s a race to the bottom, and the weapon of choice is automated content generation.

This low-effort content, which I’ve started calling “AI slop,” is designed for search engines, not for people. Its goal is to rank for specific keywords, attract a visitor, and serve them an ad. Whether the visitor actually learns something useful is secondary. This approach clogs up search results, pushing down thoughtful, human-written articles from experts or enthusiasts who took the time to share real knowledge.

Beyond being merely annoying, this AI slop can be actively harmful. AI models are notorious for “hallucinating”—confidently stating facts that are completely wrong. An AI-generated article on personal finance might invent a tax law. A travel guide might recommend a restaurant that closed a decade ago. A health article might offer dangerous medical advice. Because the text is written in a plausible, often authoritative tone, it’s easy for a reader to accept these falsehoods as fact. The sheer volume of this content makes the internet a less reliable place for everyone.

My personal original beat on this is realizing the fundamental economic mismatch at play. The marginal cost of producing one more article dropped to zero overnight. However, the potential revenue from a single ad click on that article did not. This imbalance created a massive incentive for arbitrage—generate infinite content for free, capture a fraction of a cent per page view, and scale it across thousands of domains. The web is now choking on the output of that equation.

How can you spot AI-generated content?

Since AI detection tools are unreliable at best, your best defense is your own critical eye. I’ve gotten pretty good at sensing when I’m reading something written by a machine. It’s like developing a taste for good food—once you know what to look for, the cheap stuff becomes obvious. Here are the tells I look for:

The Vibe Check: This is the first and most important test. Does the writing have a pulse? Is there any personality, humor, or opinion? AI-generated text often has a very smooth, formal, and sterile feel. It uses perfectly correct grammar but has a monotonous rhythm. Sentences are often of similar length and structure. If it reads like a slightly-too-helpful corporate HR document, it’s probably AI.

Repetitive Phrases and Concepts: AI models often repeat the same point using slightly different words, a trick to pad out the word count and hit keywords. You might see a paragraph that says, “It’s important to stay hydrated.” The next paragraph might say, “Ensuring you drink enough water is crucial.” A human writer would just move on. Look for this kind of recursive, circular language.

Generic, Actionable-Sounding Advice: AI slop is filled with advice that sounds helpful but is utterly generic. For example, an article on productivity might tell you to “prioritize your tasks,” “take regular breaks,” and “create a to-do list.” It won’t offer a specific method like the Pomodoro Technique, mention a useful app by name, or share a personal story of how the author struggled with procrastination. The content lacks specifics.

Strange Word Choices: Sometimes AI models use words that are technically correct but contextually bizarre. A human talking about a new laptop might say it has a “great screen.” An AI might describe its “luminous display panel.” In a recipe for cookies, a human might say “bake until golden brown,” while an AI might suggest you “bake until the Maillard reaction is visibly complete.” These slightly-off phrases are a huge red flag.

The Author and the Outlet: Always check who wrote the article and where it’s published. Is the author a real person with a bio, a social media presence, and a history of writing on the topic? Or is it “Admin” or “Guest Writer”? Is the website a reputable publication, or is it a generic domain like “UltimateTechReviewHub.net”? A lack of transparency about the source is a strong signal that the content isn’t meant to be trusted.

What should you do to keep your information diet healthy?

Simply getting better at spotting AI slop isn’t enough. You have to actively change how you find and consume information. The age of casually Googling everything and trusting the top results is over. Here’s how I’m adjusting my own habits:

Curate Your Sources: Instead of starting with a broad search, go directly to sources you already trust. Build a list of favorite blogs, news sites, and individual writers. Subscribe to their newsletters. Use an RSS reader like Feedly to create your own personalized feed of high-quality content. This flips the model from passive discovery to active curation.

Rely on Human Curation: Find people who are good filters. On platforms like X (formerly Twitter), Bluesky, or even in niche Reddit communities, you can follow experts who share and comment on interesting developments in their field. They are doing the hard work of sifting through the noise for you. A link shared by a trusted person is worth a hundred top search results.

Pay for Quality: This is a big one. The reason AI slop exists is to make money from ads. One of the most powerful things you can do to support high-quality, human-made content is to pay for it. Subscribe to a newspaper, a magazine, or a Substack newsletter you love. When you directly fund creators, you support an ecosystem where quality is the product, not your attention.

Refine Your Search Tactics: When you do have to use a general search engine, get more specific. Use long, detailed search queries. And here’s my favorite trick: append site:reddit.com or site:news.ycombinator.com to your search. This limits the results to discussions by real people, which are often far more insightful and authentic than a content-farmed article.

Is all AI-generated content bad?

No, and it’s important to make a distinction. AI can be an incredible tool for writers and creators when used thoughtfully. I use AI tools myself to brainstorm ideas, summarize long reports, or suggest alternative phrasings when I’m stuck. This is what I’d call “AI-assisted” content—a human is still in the driver’s seat, making all the critical decisions, and is ultimately responsible for the final output.

The problem is fully “AI-generated” content, where the machine does 99% of the work and a human does little more than click “publish.” This is the slop. The difference is intent. Is the AI being used as a bicycle for the mind to help a human create something better? Or is it being used as a factory to churn out content for algorithms? The former enhances human creativity, while the latter devalues it. The content I’m worried about is the factory-made stuff, and right now, the factory is running 24/7.

FAQ

What is “AI slop”? AI slop is a term for low-quality, often inaccurate, and soulless content created by AI models. Its primary purpose is to rank on search engines to generate ad revenue, not to inform or engage a human reader.

Can’t AI detectors solve this problem? Not reliably. AI detectors are often inaccurate and easy to fool with simple editing. As AI models get more sophisticated, telling machine-generated text from human-written text becomes harder. It’s better to build your own critical reading skills.

Does this mean the internet is useless now? No, but it does mean you need to be more intentional about how you find information. The open, easily discoverable web is getting much noisier. Your strategy should shift toward relying more on curated sources, trusted brands, and direct recommendations from people you trust.

Is followmy.ai written by AI? No. We use AI as a tool for brainstorming and research, but every article is written, edited, and fact-checked by a real person. We believe the human perspective—the opinions, experiences, and voice—is what makes an article worth reading.