Sterile Soulless Writing

AI and chatbots have been the subjects of much debate, and many articles have been written – some in favor of using them, others speaking against it.

It can be difficult to settle on a clear opinion, as the overall aspects are blurry – anything can happen tomorrow that we couldn’t predict today. And even what is already happening may be unclear or unknown to us.

AI has entered several spaces in our lives, and at times we are not thinking about it as AI – at other times, it may not by a strict definition actually be AI, but still: it is about the machines taking over some of our thinking.

Recently, some thoughts aired in a discussion made me think about the situation we are in as writers, and I actually made up my mind on some details. I may “make it down” again later, who knows, but for now, I think that I understand what I have been feeling for a while.

A beginning

A year ago, I wanted to write a series of articles about different uses of AI – try to uncover the good uses and the bad. And proofreading in combination with writing suggestions in a larger scale was one such use.

In the planning of this, however, I got a feeling of something not being right and developed an aversion against the concept and never wrote the articles. Now, just a short while later, my feeling has turned into knowledge: I see how the world is being filled with AI generated texts that the “writers” claim they just got some suggestions for from the AI.

Its extent

I see such texts also as a publication owner at Medium, where people occasionally send in texts that are clearly AI-made. And I cannot make myself publish it, even though I want to give room for all kinds of thoughts. Because, somehow this kind of stuff doesn’t feel like anybody’s thoughts. It is perfectly correct in its spelling and grammar, and what it says is mostly well phrased and nicely written – but it has no soul. There is nobody with doubt or opinions behind it. It is sterile.

There is a line of features in the tools we all use for writing, stretching from highlighting typos to fully writing the text, with everything in between covered. They can all be useful at times, and I guess that we wouldn’t want to be without, for example, the spell checker in a word processing program.

What I see, however, is a tendency for people to walk along that line and let the tools do more and more of the writing for them – still claiming that it was their text. Where we got a bit of spelling help a few years ago, we can now deliver a text that we hardly wrote any part of.

Without a soul

You can use various automated tools to help you with your writing, but that makes the text less yours, more the tool’s. And it is soulless. The part you wrote may display your unique personal attitude towards the topic, but if that is only a minor part – what is then left for the reader to enjoy?

The soulless AI-made text is kind of a synthesis of the last several years worth of texts, run through a series of filters to make it family-friendly, non-offensive, easily digestible to anyone, and free of anything that could be new and thought-provoking.

My take is that if you write an ingredients list for a food package, you don’t want to put your soul into it – so, by all means, let it happen by automatics.

But if you write something that should bring your thoughts and spirit to display, then the machines will only help to ruin those efforts – they will take the value in exchange for convenience. Take your soul in exchange for correct spelling and grammar.

Outlines

When writing longer texts, I like rich outlines with text examples and explanations of what goes into the different suggested sections. It has been suggested that such outlines can be made by the AI, and then you have a basis to discuss from with your customer – the one to receive your text when it is ready.

But how is this supposed to work? The machine invents something that you will then defend? Or, the machine says one thing, and you can then step in to renounce it and say something better?

In my opinion, the outline is for your thoughts to be displayed. This is your chance to shine.

As a technical writer working freelance for a customer, I recently presented an outline, but I didn’t use AI at all, I used my experience and some business analysis to get to the suggestions and descriptions. In this case, the customer is not used to working with writers, and I felt that I had to tell more than just a suggested set of headlines, so there were explanations and drip-wise copy-paste of parts of their old texts, plus some amount of new text that I had written to illustrate what could fit into each section.

The only automatics I put into it was Word’s spell-checker, and even that was sparsely used, as it actually makes some strange suggestions at times – especially in the technical domain where it doesn’t know much that could be helpful.

Real writing being detected as AI

It has been suggested that avoiding AI in the writing process will not help you – AI-checkers may still claim that your text is “probably AI” or similar, even if it isn’t. Speculations have been made about the nature of this, suggesting such as using Grammarly for improving the text will contaminate it with AI, so to speak, as using Grammarly will make your text sound like written by a machine.

My bet is that if you write on the basis of a dialogue with the customer, you will not sound like an AI. If you bring some real answers to real questions into play, then your text will end up being mostly unique.

Still, of course, an AI-checker could decide that it looks like something an AI could have written, especially for the often used phrases that inevitably will be found in almost every text, but I doubt that it will happen to large parts of your text.

Also – should the fact that you may, falsely, be accused of using AI be a reason for doing exactly that? That wouldn’t make much sense, would it?

Btw. – this 100% human-written text was put into GPTZero as a test, and the verdict was that the introduction was uncertain but with 62% probability of being human, the rest was human with 100% probability – until I added this paragraph, which made it 90% for the Outlines section.

Suggestion

If you want to use AI in your creative writing of any kind, the writing where you want to show your skills and knowledge, your thoughts, ideas, and opinions, your doubts, dislikes, and preferences – then use it only as you would use an encyclopedia or a thesaurus:

Look up things, read the answers, get inspired to do further searches for words or meanings using different sources, but don’t write any of it directly – use all you learned from the sessions with AI in a delicate blend with all that you learned from elsewhere to shape some thoughts of your own.

Then put all the sources away and write down these thoughts – to be enjoyed by your readers.

Photo by Carl Tronders on Unsplash

This article was originally published on Medium, some time during 2023–2024, but I have no longer any record of exactly when it was. It was planned for Substack toward the end of 2024 but was for some reason never published there. However, I did have the date for it’s planned publication on Substack, which I have then used as the original publication date for the article.

A bit complicated, I know. What is worth considering, is that when I wrote this, there was still a debate about using AI or not for writing articles. It was kind of a battle between the human and the machine, and the human lost.

As we now write 31 Dec 2025, everybody seems to have fallen for the temptation, and almost nobody still have the idea that articles and novels, and other intellectual pieces, really should be written 100% by you, not by your machines.

So, in just a couple of years, this debate has ended, and an article like this one has become historical. I brought it into All of Life XL anyway, because I think that thinking and reasoning doesn’t age. It is always interesting to look back to see which arguments were on the table, no matter the outcome. We can learn from it, I believe.