Today, I had to write a blog post.
I woke up, made a lunchbox, and dropped off the kid at school with a whole two minutes to spare (almost late, but not late). I drove back home and remembered I had to write a blog post.
But I’d just gotten back, so I needed breakfast and maybe some coffee. I also needed to feed the husband. So I got myself something to eat, fed the husband and myself, and then remembered I had a call with a potential client and a couple of emails to respond to.
And the blog had to wait.
I ended the call, and lunch came around. I remembered I needed to write a blog post, so I started going through my content bank and the ideas scribbled in my notes while cooking. Then that one needy client (you know which one) called. Lunch didn’t burn, and we ate, but there was still no blog post.
Lunch passed. I had two more calls, 20 more follow-ups, and apparently not enough time to write one silly blog post.
And then it clicked. Maybe the endless supply of AI-generated content, with its “In today’s digital landscape” openings and endless “it’s not this, it’s that” comparisons, isn’t just a sign that people are lazy.
Maybe we’re all just busy.
AI-Generated Blog Posts
Yes, you and I know they exist. We’ve all read at least one article or piece of content and thought, THIS HAS TO BE AI… Sometimes it’s the EM dashes — cute but overused — or the verbiage… sometimes it’s the same predictable structure, the same phrases, the same arguments we’ve seen a hundred times. And soon, even the Claude watermark might become another thing people look for.
But as much as we all love to point to this s, thing, none of those signs, on their own, prove who wrote something. A human can use EM dashes (actually, I used to LOVE using them since I was a teenager who read tons of books and saw them CONSTANTLY in everything I read), repeat familiar phrases, and write in a predictable structure… and you can definitely ask AI to do the opposite or to “humanize” your text.
Still, there are reasons AI-generated content has become such a big part of the conversation.
According to research published by Graphite in May 2026, approximately 49.9% of sampled English-language articles published online in the first quarter of 2026 were classified as primarily AI-generated. The study analyzed 55,400 URLs from Common Crawl, using three AI detectors and averaging their results.
That’s a significant share. But let’s be clear about what it means: these were articles in a particular sample, classified using detection tools. It doesn’t mean half of everything on the internet was written by AI, and it doesn’t tell us how every individual article was created.
On the other hand, a study by researchers from Stanford University, Imperial College London, and the Internet Archive estimated that about 35% of newly published websites (not articles, but full websites) were AI-generated or AI-assisted by mid-2025. Different study, different sample and definition, still pointing to the growing presence of AI-assisted writing online.
But does more AI-generated content automatically mean more useful content? Or better content? Or content people actually find?
Not necessarily.
Graphite’s separate research into search visibility found that articles classified as primarily AI-generated appeared infrequently in Google and ChatGPT results. That suggests publishing more AI-generated articles doesn’t automatically translate into visibility. It does not, however, prove that most AI-generated content fails to rank or that the use of AI itself is the reason a page doesn’t appear.
And then there’s trust.
The Reuters Institute’s 2026 Digital News Report found that global trust in news stood at 37%. Its findings also showed that only 22% of respondents trusted news on social media and 20% trusted news from AI chatbots. And while these figures are about news, not marketing blogs, they are proof that credibility matters.
So, yes, AI-generated content is everywhere. And yes, people have reasons to question what they read, but that doesn’t mean every piece of content that looks like a robot wrote it is AI: AI is behind everything, but not everything is AI.
Can You Actually Tell If Content Was Written by AI?
Now, to all those numbers, there’s a disclaimer we need to add: AI detectors are imperfect.
Even research that uses multiple detection tools is still relying on those tools to classify content. Different tools can produce different results, and the conclusions depend on the sample, the methodology, and what the researchers mean by “AI-generated.”
That doesn’t make the research useless. It means we need to read the findings carefully rather than treat an estimate as an unquestionable fact.
And the problem isn’t just whether a detector gets the answer wrong. It’s who gets flagged.
According to a 2023 study highlighted by Stanford researchers, seven of the most popular AI detectors classified approximately 61% of essays written by non-native English speakers in a particular test set as AI-generated, even though they were NOT written using AI… English as a second language, much?
Why? The researchers found that these tools could penalize writing with less linguistic variation and more predictable word choices. Those patterns can appear in writing by people who speak English as a second language, just as they can appear in AI-generated text (HELLOOOO!! that is us. Nice to meet you).
And yes, we know that said study is from 2023, and it is not proof or an indicator that every detector today has the same error rate. But it does demonstrate why a detector score should not automatically be treated as proof of authorship.
And this isn’t just an academic problem.
Imagine a business owner who writes their own website copy, a student submitting an essay, or a marketer putting together a blog post between client calls. If a tool flags their writing, does that mean they used AI? No. It means the tool identified patterns it associates with AI-generated text.
Those are two very different things.
And Back to Being Busy
It is almost 8:00 PM, and this article is still in progress. I still have a ton of things to do, and I just got another call… And honestly? The easiest answer would be to prompt AI to write the post instead of researching the topic myself, reading through sources, deciding which numbers to include, and typing every single word.
Because even if I typed every word myself, I’m a tired, over-scheduled human putting something together at the last minute, using simple, clean, to-the-point ideas between client calls and emails, a damn AI detector could still read it as a robot.
And if a robot writes something messy on purpose, it might pass as me.
Probably the point here, besides how easy it is to simply prompt your next piece of content and continue with life as usual, is that we can’t reliably identify authorship just by guessing. And that guessing game isn’t helping anyone.
A detector score cannot see the lunchbox, the two-minute school drop-off, or that one client who called right when lunch was about to burn. It only sees patterns.
And here’s where I think this matters most for marketing: content is meant to communicate something, not be perfect. A blog post can be written entirely by a human and still be generic, inaccurate, or completely useless to its audience. AI-assisted content can also be accurate, thoughtful, and genuinely helpful.
The more useful questions are whether the information is accurate, whether the content offers value, whether it sounds like the brand it represents, and whether it gives the reader a reason to care: a specific detail, a strong opinion, or a recognizable voice can make writing more engaging, but none of those things proves a human wrote it.
What matters is the thinking behind the content, the care taken to verify it, and the value it delivers to the reader… and that’s where we fall back into one of our favorite phrases: AI is meant to be a tool, not a replacement.
So the next time you read something and think, THIS HAS TO BE AI, give it a second. It could be a robot. It could also be a tired human at 8:00 PM with cold coffee, doing the best they can between calls.
As for this post, I’ll leave you with the same doubt you started with: Who wrote it? Let me know what you decide.
