This post is my personal contribution to reducing AI-writing complaints on the TL.

People know when your “thought leader” posts were written by AI. In fact, as a heavy AI user/AI-savvy person (and there are many of us), I can tell when you didn’t even bother to reduce the AI-ness of what was generated. We’re tired of such a lazy approach to communication. You need to do better, so we can actually assimilate and enjoy reading from you.

These days, I automatically scroll past any post that is obviously AI-generated, which is easy to spot after the first two sentences. It doesn’t matter what you’re trying to convey. The scrolling away is an auto-response. In fact, it takes an immense amount of mental power to turn off that aversion and read the post. I literally have to convince myself that it is a case of ignoring the packaging and taking the content.

As someone who still uses AI to generate portions of my posts, at risk of sounding hypocritical, allow me to slide in a bit of personal history. In 2023, I found myself writing more after a long hiatus. My new venture into blogging necessitated this. That’s how I found out I had become very bad at writing. For context, in high school, I was the science student who bested all the art students in literature reviews and essay writing. My essay writing was top-notch, thanks to my raids on my dad’s library, where I had the privilege of studying with books like Effective English by Michael Montgomery et al. Studying with books like that and reading poetry by older writers made Intensive English (the government-approved textbook) seem like a watered-down experience.

In summary, I wrote so much better than everyone at my level. Even our English Language teacher suspected that someone else was writing my essays for me. After many trips to the staff room to prove my innocence, including testimonials from other teachers, she was eventually satisfied.

I gave this background information to give an insight into the kind of writer I was growing up. Sometimes I read my journals from back then, and wonder if I’m the same person who wrote like that. Blogging exposed my writing lapses, and like everyone, I chose the easy way out by fully generating my posts with AI. Some of my old posts on syntaxandscript.com reflect that (I try to edit and update them).

Last year, I started noticing my disdain for the AI slop that is everywhere you turn on social media these days. AI slop’s definition in this sense, being posts (especially texts) that are obviously AI-generated with zero edits, thereby excluding a chunk of people from engaging with the content, regardless of its importance.

A lot of factors contributed to my growing disdain. First is my activity on Reddit, which, coincidentally, increased in the same year. I spent long research hours on subreddits that reinforced my beliefs. But I was still using AI to generate large portions of my posts (how hypocritical!). In my defense, I was fluffing out parts of my posts that I couldn’t hack on my own. Even as I type these words, I feel the urge to “better structure” this post with Claude. But I’ll let the unstructuredness of it speak for the spontaneity of the post and the urgency of its action points.

One of the lessons I took away from being on those subreddits is that I needed to do better for myself while using AI tools more strategically. Essentially, leveraging AI  to compete as a creator in this age while maintaining my critical thinking and writing skills. This awareness has resolved my writing and research struggles by just doing the thing and letting AI be the last resort.

These days, maybe because of how vocal I have become about my stance on AI-generated content, my TL across all my socials is littered with AI-disdain posts. What no one has really said is how to unAI your AI-generated posts. This post is my twopence that doesn’t involve using another AI tool to “humanize” your writing. You can definitely do that if you:

1. Want the lazy approach 
2. Work as a writer who churns out dozens of posts per day. I don’t expect you to reduce your productivity, and as a result, your wage, because of what someone online said.

That makes this post something you’d find valuable if you genuinely want to write better than you currently do, while still leveraging AI tools in your writing.

I’m going to start with something fundamental. When I noticed how bad my writing had gotten, the first thing I did was to search for a course on how to write better. Thankfully, I found one. Surprisingly, it was on LinkedIn Learning – Shani Raja’s Writing with Flair: How to Become an Exceptional Writer. I took this course three years ago. The fact that I’m still writing about it lets you know the impact it had on me.

Take the course and thank me later. It’s on Udemy too. It’s a paid course (~₦11,000), but it’ll be the best bucks you’ll ever spend on a course.

Implement those writing skills.

I use Claude to brainstorm my ideas and criticise my drafts. I find it incredibly useful in this aspect. By the time I’m done implementing the corrections, I have a long post like this, all written off the bone of my initial idea, without using an AI model to write a full sentence. 

In cases where I’m unable to flesh out the idea on my own, I ask Claude/ChatGPT to do it. Then, I edit.

I understand that most people can’t edit or even do it well. Shani Raja also has an editing course – Editing Mastery: How to Edit Writing to Perfection. You can also check YouTube for videos on how to write and edit better. There’s plenty there. 

Even without taking a course, there are telltale AI-speak signs in most posts you’ll encounter online today. Here’s my list:

• For social media posts, especially ChatGPT-generated ones, asides the overuse of emojis, there’s em dashes (___).

• The obvious text patterns like:

 “But here’s the thing, …”

 “It’s not this, it’s that”.

• Short sentences; sometimes one or two words in a sentence. You can’t mistake the pacing. It’s short and it’s everywhere now.

• The rule of threes: Everything it lists is in threes, even when it can be more or less.

• So many unnecessary questions that lead up to an answer. E.g: “The result? …”

• If the sloppiness (read as AI-ness) of the writing is not obvious in the body of the post, it will be in the closing.

Kindly add any of the telltale signs you know in the comments.

DISCLAIMER: Since AI models were trained on volumes of human writings, it’s inevitable that a lot of people naturally write like AI or that AI writes like a lot of people. It’s unfortunate that they’ll have to change their writing style while AI races to catch up to that as well.

There’s also the case of unconsciously developing an AI-writing style simply because that is what is prevalent in the media. You are what you consume. Please, take care in this regard. 

This post is already longer than I imagined. I’m happy to share that AI didn’t have an input in it. Which I’m ordinarily not supposed to be proud of, as a writer, but given the premise…

In a nutshell, if you do half of what is recommended in this post, there’s no way your posts won’t read better than they currently do.

TL-DR:

• People know you use AI to write. The least you can do to respect your audience is to put in a little effort in editing the post.
• Even if you know how to write now, you are at risk of losing that skill from disuse or consistently outsourcing your writing to AI.
• Take one or two writing courses to refresh your skills.
• You don’t need another AI tool to “humanize” your AI-generated posts. But do what you must.
• There are tips in the post you definitely missed out on because you couldn’t spare 7 minutes to read it. Plus, there’s a small personal gist enclosed. 😉