The internet used to have a spam problem. Now it has a sameness problem.
Product reviews that all sound alike. Blog posts that say a lot without saying anything. LinkedIn updates that read like they were written by the same enthusiastic intern. Search results stuffed with articles that technically answer your question and somehow leave you knowing less.
A growing share of what you read online was generated by AI. Some of it is fine. A lot of it is filler. And some of it is designed to mislead.
So how do you tell?
Why It Matters
AI-generated content isn't automatically bad. Plenty of human writing is bad too. The problem is scale and intent.
When it costs nothing to produce a thousand articles, the economics of the internet change. Content farms can flood search results. Scammers can produce convincing fake reviews. Brands can pump out endless posts that sound the same. I wrote about this flood in Gemini Omni Is Here and The Future of Work Is Just Machines Generating Slightly Better Nonsense.
Inside companies, it shows up as an avalanche of memos and decks that nobody reads. I called it the AI arms race of irrelevance. And in marketing, it's making brands sound interchangeable, the problem I explored in When Personalization Feels Impersonal.
Learning to spot it is becoming a basic literacy skill.
10 Telltale Signs of AI-Generated Writing
No single sign proves anything. Look for clusters.
1. It's fluent but empty. Smooth sentences, perfect grammar, no actual insight. You finish reading and can't name one specific thing you learned.
2. It hedges everything. "It's important to note." "There are many factors to consider." "Ultimately, it depends." AI tends to cover every base instead of taking a position.
3. It loves a tidy structure. Neat lists of three. Every section the same length. A conclusion that starts with "In conclusion" and repeats the introduction.
4. It uses the same pet words. Certain words and phrases show up far more in AI text than in normal writing: landscapes being "navigated," things being "crucial" or "pivotal," "delving into" topics, "unlocking" potential. One instance means nothing. Five in a paragraph is a pattern.
5. It lacks specifics. Real writers name names, cite dates, describe particular moments. AI often stays general, because specifics are where it's most likely to be wrong.
6. The specifics it does include are wrong. Fabricated statistics, misattributed quotes, studies that don't exist. If a claim sounds precise, check it. Even professionals get caught: KPMG had to pull a report over apparently hallucinated content (more here).
7. No lived experience. Human writing often includes personal detail, opinion, frustration or humor rooted in experience. AI can imitate this, but it often feels generic, like a stock photo of an emotion.
8. It answers a slightly different question. Search-bait articles often circle your query without addressing what you actually needed to know.
9. Reviews that sound identical. Clusters of reviews posted around the same time, with similar phrasing and suspiciously balanced praise, are a red flag.
10. Too perfect for the context. A casual forum reply written like a corporate press release. A personal email with zero personality.
Why AI Detectors Aren't the Answer
It's tempting to paste text into an "AI detector" and trust the score. Don't rely on it.
AI detection tools are known to produce false positives, flagging human writing as AI-generated, and false negatives, missing AI text that's been lightly edited. Non-native English speakers and people with plain, formal writing styles are especially likely to be wrongly flagged. Some students have faced real consequences because of unreliable detector results.
Use detectors as one signal at most. Your own critical reading is more reliable.
How to Protect Yourself
Favor sources with real people behind them. Named authors, a track record and a clear point of view are harder to fake at scale.
Check claims that matter. If a decision depends on a statistic or quote, find the original source.
Read reviews skeptically. Look at the distribution of ratings, the specificity of complaints and whether reviewers have a history.
Notice the "nothing" feeling. If you finish a piece and can't summarize a single concrete point, that's your answer, whoever or whatever wrote it.
Keep your own thinking sharp. The better you are at reasoning on your own, the easier it is to spot reasoning that isn't there. (See Is AI Making Us Dumber?)
For Businesses: Don't Become the Filler
If you run a business, this cuts both ways. Customers are getting better at recognizing AI filler, and they trust it less. Recruiters notice AI-generated resumes. Readers notice AI-generated blogs.
Use AI to help you work faster, but put real expertise, real opinions and real specifics into what you publish. In a world drowning in generated text, a recognizable human voice is a competitive advantage.
Let's be clear: the goal isn't to ban AI from writing. It's to make sure somebody, somewhere, still has something to say.