By Marcus Coetzee, February 2026.
I can’t stand AI writing anymore. I immediately skip over any blog post, LinkedIn post or Substack article that appears to be written or illustrated by AI. My resolve is much greater if the post shouts, “Look at me. See how clever, wise, successful or virtuous I am.” I quickly skip past these. My strict reading filters no doubt produce false positives, but that’s a risk I’m prepared to take.
This type of AI writing is easy to recognise. The short sentences are often laid out directly below each other without paragraph spaces. The large em dash. The catchy hook. The lists of short bullet points. The lack of useful detail. The overuse of certain types of phrasing such as: “It’s not just this, it’s that” or “If you think X you’re wrong, it’s Y.” The same applies to AI artwork. Once it’s seen, it can’t be unseen.
Let me clearly state. I am not a Luddite averse to using AI. It is an essential tool. I use it every day for research and for testing ideas. I use it for tasks that I would have assigned my PA or researcher when I had them working for me. AI can certainly help one’s workflow. I’ve even written two articles about AI. The first was on how AI is a disruptive tool, much like others I’ve experienced in my lifetime (e.g. computers, internet, mobile phones, Google). The second was on how AIs cannot create genuine art but merely assimilate artists’ work.
I don’t like the way that AI wants to write posts and essays, and yes, I’m aware that one can instruct AI to write in different styles. AI learned from the best copywriters. Recently, I even see writers unconsciously copying an AI writing style because they see this style gaining visibility on LinkedIn and other social media platforms. Such writing is converging on boring sameness.
So we now have a problem with AI copying writers who write in an AI style, copying writers using AI, who are copying the original successful writers. I can imagine an internet where so much stuff is being made by AI copying and learning from other AIs pretending to be human. We already have an AI arms race between AIs submitting assignments, proposals, emails and CVs versus AIs evaluating and responding to them. The original human essence and insights are lost.
A recent absurd example was a family member sending me a Gospel song to cheer me up during a low point in my life. I indulged them and gave it a listen, since I normally listen to classical music or heavy metal. Something about it seemed off, so I investigated and discovered it was all written, produced, sung and marketed by AI. This feels so wrong. I expect this is part of the enshittification of the internet.
If I want to understand something, I’ll simply use a search engine or AI to summarise it for me. I’ve learned to guard against the tendency of AIs to agree with my assumptions in our discussions. Then, if I want to know more, I’ll explore further through a book or podcast, or I’ll schedule a discussion with an expert.
If I want to read someone’s insights, I expect that these would have taken time to coalesce and bubble up through their notes and conversations until they finally manifest on paper. The type of insights I want to read aren’t those developed and written during a one-hour writing session as part of a posting schedule. The insights I want to read are deeply human, grounded in personal experience and uncontaminated by machines. They are the types of insights and personal revelations I want to reflect on and discuss with people around me.
I’m not advocating against AI to the level of Paul Kingsnorth’s proposed “Writers Against AI” movement, even though I love his work, since I experience the utility of AI daily. But I believe authors and other creatives should disclose the extent to which AI was used in their work, in much the same way as the various open source licenses disclose how intellectual property can be used.
My view is that if I want to read deeply human insights, then I’m going to prioritise those written by humans without the assistance of AI. I’m going to accept the human flaws, knowing that I’ve read something entirely human-made. But if I’m after technical or academic knowledge, then I’ll use AI myself to extract and package that knowledge first-hand. This polarised approach works well for me. Then if I still want more wisdom and insight, I’ll always end up with a good solid book in one hand and a cup of coffee in the other.
(This article was written without any AI support. Any flaws are man-made.)