Confessions of an AI Writer: The Patterns You Can't Unsee

I sat through a presentation last week where a senior colleague wrapped up by saying "Thanks ChatGPT" with a smirk.

Everyone laughed. They all knew. It was the most corporate-sounding, perfectly structured, boring presentation they'd seen all year. I couldn't get to the coffee fast enough.

That's the thing about AI writing - technically perfect, totally empty. And once you see the patterns, they can't be unseen.

Let me show you how to spot these AI tells, and I’ll pull back the curtain on how I use AI without falling into these traps.

The statistical trap: why AI makes everyone sound the same

The funny thing about an AI writing: it doesn't actually write - it predicts.

AI is designed to select the most statistically relevant next word based on patterns it's learned from training data.

Us mere mortals might say "the big red house" in one sentence, then switch it up with "the crimson toned mansion" in another. AI? It sticks to the most common version every time because seriously who is whipping out “crimson toned mansion” enough in their writing for it to be considered statistically important?

It’s getting better but that leads us into a boring predicament - we all sound the same.

The telltale signs your writing has been “AI-washed”

Some patterns I've noticed after reading a lot of AI writing:

1. Overused LLM words: AI loves to sound smart. Watch for "delve into" (instead of explore), "navigate the landscape" (instead of understand), "underscore" (instead of show), "showcase," "pivotal," "realm," and the worst offender - "meticulous." Real humans don't stack fancy words unless they're writing a PhD thesis.

2. The transition trap. Phrases like "Moreover," "Additionally," and "Furthermore" appearing every few sentences. Real humans don't write like that. We use these transitions sparingly unless you’re writing an essay for university or college.

3. The thesis statement syndrome: My absolute favorite of the telltale AI signs. A direct declarative statement that reads like a high school essay introduction. It's doing that statistical prediction thing - picking the safest and most used way to start a topic.

   So you’ll see:

   • "The importance of X cannot be overstated..."

   • "X plays a crucial role in..."

   • "In today's fast-paced world, X is essential..."

   • "X is a fundamental aspect of..."

   
   AI will announce the topic before actually discussing it. Real humans, well we’re much more likely to dive in with something engaging like a story, a question, or an observation.

How I use AI without losing my voice

I’m not shy in saying that I use AI to write - I use it to write this newsletter, improve emails, and produce better technical documentation for my clients better and faster than I could ever produce alone.

But I don’t simply slap “write me a newsletter” into Claude, there’s a process:

• I start with a topic: Usually from a boring meeting or a conversation about AI or, if I'm dry on ideas, from my RSS feeds. Then I ask Claude about its real-world impact at work:

• I ask Perplexity: to find me authority pieces on the validated topic. This is where I get statistics like those above. It also verifies that the topic I’m writing is again relevant to our worklife as the companies I restrict my Perplexity search to have conducted research which backs the topic. If there’s no tangible evidence, there is likely no real-world outcome for you.

• I decide on the goals and create a rough outline: my goal for this edition is to dive head-first into a taboo topic, remove the taboo, and pull back the curtain on how I do things.

  This week went like this:

  • AI writing is here to stay so we can either complain about it or deal with it

  • How to tell if something is written by AI and why the result is usually pretty boring therefore you should not rely on it solely

  • Show people how I use AI to maximise the amount and quality of what I write resulting in better information for them

  • I then start writing a rubbish draft full of notes to myself, mistakes, and arguments that aren’t rounded.

• I talk polish with Claude: spelling, grammar, structure. We discuss how well rounded the argument is and if I’m achieving the goal of leaving readers with something tangible and usable in their workday. I ask it to give me a percentage on how I ranked against the goals of the article and I am looking for 100%. If I’m not achieving the goal to 100% what’s the point of writing the newsletter?

The bottom line

That statement above: what’s the point? is key.

I could go ahead and write unassisted leaving gaps, forming lacklustre arguments, leaving infuriating grammar and punctuation errors, and ultimately wasting your time. But what’s the point of that?

We’re all time-poor so how do we get the point across quickly and efficiently, that is the role AI plays in our world.

But efficiency doesn't mean sacrificing authenticity. The key is using AI to amplify your voice, not replace it.

Because in a world where AI makes everyone sound the same, your unique perspective is more valuable than ever.

IN PARTNERSHIP WITH THE RUNDOWN AI

Learn AI in 5 minutes a day

This is the easiest way for a busy person wanting to learn AI in as little time as possible:

1. Sign up for The Rundown AI newsletter

2. They send you 5-minute email updates on the latest AI news and how to use it

3. You learn how to become 2x more productive by leveraging AI

Sign up to start learning.

FACTS TO IMPRESS PEOPLE AT PARTIES

In 2018, an AI called a Generative Adversarial Network (GAN) created a painting that sold for $432,500.

My question is, ethically, who should get paid? The artist?

Catch you next week!