I saw a post from The Digital Picnic on Instagram this week about LinkedIn making changes to prioritise more authentic, human content, and my first thought was ... thank goodness.
When did everything become content?
For a while now, I’ve been finding LinkedIn an increasingly difficult space to spend time in, to the point where now I just don’t. There’s so much content being produced just because someone has decided they need to post every day, and so much of it feels like a substance-free variation of something I’ve seen or read a hundred times before.
Everyone has a course to sell, advice to give, or some remarkable business revelation they’ve managed to pull out of an otherwise ordinary everyday experience and turn into a profound lesson on leadership. I’ve found myself scrolling past most of it, which is a shame because I genuinely enjoy stories and storytelling, and I love hearing what people have to say, particularly when they’ve got an interesting take or have experienced something that makes me stop and reconsider my own thinking.
So when I heard this, I went looking into what LinkedIn was actually doing, and it turns out they’ve been working on this for a while. In September, LinkedIn announced that its efforts to reduce what it calls ‘AI slop’ had resulted in a 40% reduction in views of content classified that way. More than a million members had also used its new ‘Seems like AI slop’ reporting feature within the first two weeks of its introduction. They’re not banning AI-generated writing, which I think is an important distinction, but they are trying to reduce the amount of generic, repetitive content that contributes very little to the conversation.
And I’m here for that.
But who decides what’s human?
I’ve also been having another conversation over on Substack, about AI detection and what happens when we start allowing technology to decide whether something has been written by a human. Because while I’m on board with platforms trying to reduce the endless stream of generic content, I’m increasingly concerned about how we’re identifying it and what that means for the people who actually create for a living.
After seeing the Substack collaboration with Pangram, I started putting some of my own articles through its AI detector, partly out of curiosity and partly because I’d been seeing more conversations about writers having their work questioned. I wanted to see it for myself, so I tried Pangram first and then a few other detection tools to compare the results.
One of my articles came back with a particularly high AI score, which was interesting considering I’d written it more than twelve years ago, well before ChatGPT and generative AI became part of our everyday toolkit. I then started testing other pieces of writing across different detectors, and the results were all over the place. One would tell me the writing was entirely human, another would insist the same piece was AI-generated. Another put it somewhere in the middle.
I shared some of this on Substack, and the conversation that followed made me realise how many other writers are experiencing the same thing. People who have spent hours developing their work are finding themselves having to defend whether they actually wrote it. And I find that extraordinary, particularly when you consider that we’re talking about software making assessments about something as individual as a person’s writing style.
And it’s not just anecdotal, either. Research published in September 2026 in the journal AI and Ethics examined the reliability of AI detection tools, including an experiment where the same human-written manuscript was put through five commercial detectors on the same day. The results ranged from ‘0% human’ to ‘Human Generated’. The exact same piece of writing, tested on the same day, with completely contradictory conclusions. Sound familiar?
And just this month, a systematic review of 90 studies published in Discover Education found that false-positive results continue to vary significantly between detection tools, with researchers raising concerns about what happens when those results are treated as evidence of authorship. So despite how quickly this technology is developing, we’re still having conversations about whether we can reliably identify who wrote something. And yet people are already being asked to defend their work against these assessments. How have we reached a point where a piece of software can cast doubt on someone’s writing, but the writer is the one expected to prove it wrong?
Now, to be fair, LinkedIn’s systems for identifying low-quality content aren’t necessarily the same as the commercial AI detectors I’ve been experimenting with, and LinkedIn has said that it’s looking for generic content lacking genuine perspective, rather than simply punishing people for using AI. In fact, LinkedIn explicitly acknowledges that AI can be useful for helping people write, provided the ideas and perspectives remain their own.
Which is a distinction I agree with.
I’m not anti-AI. Far from it. I use it regularly in my business, and I’ve never felt the need to pretend otherwise. I use it to research, to question ideas, to have conversations about things I’m trying to understand and sometimes to work through an argument before I sit down to write. It’s much the same way I’d talk something through with my husband or another writer, except there’s someone available to bounce ideas off at whatever ridiculous hour I happen to be thinking about something. I don’t believe using a tool to help develop an idea automatically makes the resulting work any less human. What matters to me is whether there’s actually a person behind it with something they want to say.
Are we changing how we write to prove we’re human?
And perhaps that’s where I’m getting frustrated with this whole conversation, because somehow we’ve ended up in a situation where writers are being encouraged to produce original, thoughtful, human content, while simultaneously being made to worry that their natural writing style might be mistaken for AI. We’re expected to be visible, to publish regularly, to understand algorithms, to build audiences and to make sure everything we create is sufficiently engaging to earn a few seconds of someone’s attention. And now, on top of all that, we’re wondering whether we need to change the way we write to convince a machine that we’re human.
How ridiculous is that?
I’ve caught myself thinking about changing sentences because they might sound too much like AI, and then I’ve wondered why on earth I’m doing that. These are my words. This is how I write. I’ve been writing long before any of these tools existed, and yet here I am considering whether I should make my writing less like my writing to satisfy something that can’t reliably tell me who wrote it in the first place.
And I wonder what that does to creativity over time.
We already have an enormous amount of pressure to turn everything into content. Every experience becomes an opportunity for a post, every thought needs a hook, and every piece of writing is measured against how many people saw it, liked it or shared it. There’s very little room for exploring an idea, getting something wrong, changing your mind or writing something that doesn’t have a particularly useful conclusion. Sometimes I want to write because something has been bothering me, or because I’ve had an interesting conversation, or because I’ve noticed something about the world that I want to understand a little better. I don’t always have my top five lessons to share at the end of it.
I just want to write.
And I think that’s what worries me most about where we’re heading. Not AI itself, because I genuinely believe it has a place, but the possibility that we’re becoming so concerned with how our work will be received, distributed and judged that we’re starting to lose the ability to authentically create it.
So what’s the answer?
I suppose I’m coming around to the idea that we need to stop giving algorithms and detection tools quite so much authority over what we create. Keep researching, keep questioning, use the tools that help, ignore the ones that don’t, and write things because you have something you want to say rather than because a platform has decided it’s time for you to publish again.
And if the algorithm doesn’t like it, or an AI detector decides I’ve managed to write something that sounds too much like a machine, maybe the answer is just to say fuck it and keep writing anyway.
I’d genuinely love to hear how other writers and creatives are approaching this, particularly those who’ve had their own work flagged as AI. Are you finding yourself changing the way you write, or have you reached the point where you’re just going to create what you want and let the algorithms do whatever they’re going to do?
Because I’d rather spend my time becoming a better writer than trying to become a more detectable human.
Originally posted on LinkedIn here
If this struck a chord, Richard and I explore these sorts of questions every week on the Richly Told podcast. Business, family, travel, leadership and all the complicated bits in between.
New episodes land weekly on Spotify and Substack.