The truth is that being able to spot AI giveaways in writing is becoming a really important skill. In fact, being able to identify the telltale signs of AI in any capacity is only going to get harder and more important as time goes by.

As a content writer and SEO strategist, I come across AI writing on a daily basis, and as a result, my AI-senses are razor sharp. There are so many AI giveaways in writing that I can spot a mile off, but not all of them are as obvious to the average reader. That’s why I’m sharing 10 telltale signs of AI for you to watch out for in the future.

Or, let’s be honest, how you can edit AI writing to pass as human.

• Using AI as a Freelance Writer
  • DO use AI…
  • DON’T use AI…
• 10 Obvious (and not so obvious) AI Giveaways
• Becoming a Legitimate Content Writer

Using AI as a Freelance Writer

I’m not here to side-step reality. These days, pretty much everyone is using AI in some capacity. And I’m not against this. AI can be a really helpful tool for freelance writers, with so many use cases that are both legitimate and moral.

Let me be clear, I am definitely not telling you to copy and paste a brief into ChatGPT and submit the answer. That’s a stupid idea for way too many reasons to list. You can, however, use it to assist in your writing process from ideas to research and proofreading.

What does really irritate me is when I’m working with other writers’ work and see the way too evident signs of AI writing everywhere. This is giving us all a bad name. It’s also just lazy and boring. Even in a personal capacity, I get annoyed by AI writing because I know it isn’t going to provide nuanced, interesting, and unique information.

Here’s my quick cheatsheet on when to use AI as a content writer:

DO use AI…

• As a research tool to learn more about specific topics
• During concept generation and for bulk ideas when brainstorming
• For super short formulaic copy first drafts, such as social media captions and product descriptions

DON’T use AI…

• Without reading over and editing text to be personalised and fit for purpose
• To write long-form content for professional or school submission
• If you don’t have express permission from your employer/school/publisher

10 Obvious (and not so obvious) AI Giveaways

1. The Em Dash

Oh, the em dash…probably the most renowned AI giveaway. The em dash (—) is a long dash used by writers to add emphasis, extra information, or create a pause for dramatic effect. The problem? AI bots love it. Supposedly, they use it so frequently as it helps text to sound more human…ironic, right?

I’m not telling you to never use the em dash, but it should be used carefully. The key is to not overuse it. Instead, keep sentences interesting with varying punctuation, e.g., ellipses and brackets.

2. Bold Words and Phrases

Have you heard the SEO tip to bold certain words and phrases to help on-page readability? Well, so has AI. If you ask an AI model to produce longer-form paragraphs or text, it’ll often erratically bold key words and phrases.

This then becomes a huge telltale sign that the writer simply copied and pasted the text when the AI’s bolding doesn’t match the rest of the text. If random words and phrases are bolded in just one paragraph…it’s pretty clear which one the writer didn’t actually write.

3. Change in Spelling

Most AI models are programmed to use American English with certain default grammar rules, such as whether or not to use the Oxford comma and things like accents on letters. If I’m reading a post which suddenly switches from British to American English, it’s pretty obvious to me that the writer has used AI for that paragraph.

Something I come across a lot, particularly living in Japan, is that blog posts and web content will be inconsistent in the translation of words. A super common example is Tōkyo vs Tokyo, with the ō appearing and disappearing depending on whether the sentence is human-written or not.

4. Shallow Points

You have to remember that AI is not actually thinking. They are mathematically breaking down a prompt and putting together sentences based on the huge amount of texts that the model learnt from. What I’ve found that to mean is that the prompt results are typically not very deep or complex. AI tends to smooth over knowledge gaps with broad wording, employing generalisations instead of concrete details.

As a result, ends aren’t really tied together. There’s usually little to no nuance to points, and paragraphs answer one question…and one question only. Great writing is complex. It can tackle multiple ideas at once, drawing together differing concepts into one conclusion.

Read between the lines of a piece of writing. Is there subtext? Does the deeper meaning actually make sense? And, does it seem like the writer had a genuine and deep-rooted comprehension of the subject?

5. The Summary Sentence

This one really bugs me. AI just loves to end each paragraph with a summary sentence, and usually one that begins with whether, e.g., “Whether you’re looking for sun, snow, or city, Japan has something for everyone.”. Trust me, if a paragraph ends with this sentence pattern, it was not human-made.

In normal writing, there’s no need to end every paragraph with a summary…I’ve just read the information in three sentences and don’t need an immediate reminder. AI, on the other hand, is designed to prioritise clarity to a fault, hence the easy spot.

6. Rule of Three

Now, I’m not here to bash on the rule of three. We’re taught it in school for a reason, and I genuinely do believe that listing things in threes is how the human brain works. However, as describing things in threes is so prevalent in the writing AI models are trained on, they’ve gone a little far with the concept.

A list of three here and there is great, but the rule of three popping up every other sentence definitely smells a little fishy. It’s the frequency and predictability of its use that can feel artificial, so, if anything, this is just generally good writing advice to mix things up.

7. Clichés

AI just loves a cliché. And why wouldn’t it? When things are used a lot in human writing, AI assumes that using them will make them seem more human-like. Once again, avoiding this potential AI telltale might actually just work to make you a better writer.

Here’s a quick list of the clichés I immediately suspect are AI-written:

• A hidden gem
• Off the beaten path
• Old world charm
• A feast for the senses
• Nestled in the heart of…
• The perfect blend of old and new

Clichés like this sound descriptive, but don’t actually provide any real detail. AI often uses these because they’re safe, familiar, and can apply to almost anything. The problem is they’re vague and interchangeable, which makes the writing feel generic and AI-like.

8. The Phrase “Think…”

This one usually combines a bunch of giveaways, including the em dash, rule of three, and clichés, for example:

“Kyoto blends old-world charm with slow, meaningful experiences—think private tea ceremonies in hidden machiya houses, morning walks through bamboo groves, and riverside dinners.”.

For me, this means I just rarely use this sentence pattern any more. There’s nothing technically wrong with it, but if it is common enough for AI to use it all the time, I’m not interested in often including it in my writing.

9. No Negativity

I think this is something more and more people are starting to see. In essence, AI bots are programmed to provide positive-leaning answers. I literally asked ChatGPT why this is, and got this answer:

“I aim for optimistic realism: useful information presented in a way that motivates and empowers rather than overwhelms.”.

In other words, positive reinforcement tends to keep users coming back more than negative. Now, AI models can use whatever retention techniques they want, but this lack of negativity has become a clear AI giveaway in writing. Most topics are not all good. Applying critical thought and even some subjectivity is important in crafting a well-written piece. So, next time you notice a lack of nuanced balancing of good and bad, be aware that you might be reading AI content.

10. Man-Splaining

I know that AI isn’t any gender (and interestingly, when I asked ChatGPT which gender they present as more, they said female), but that doesn’t mean it isn’t man-splaining. What I mean by this is that AI models have a tendency to over-explain the obvious, e.g. “Tokyo, the capital of Japan, is a bustling city…”. You don’t say.

I do think this is another example of how AI writing lacks specifics and the true human experience, e.g., most people already know what Tokyo is. It’s also prioritising clarity once again…but to a fault. My advice? Don’t waste valuable page space repeating inferred information. In this attention-hungry world, every line of text matters, and each sentence needs to bring a new concept to the reader.

Becoming a Legitimate Content Writer

If you want to take becoming a content writer seriously, then you’re in the right place. I share my experience, advice, and guides on freelance writing every week to help you land your first paid client.

Check out some more content writer resources below: