7 Words That Seem Like AI Content (But Might Not Be)

Delve. Tapestry. Testament. 

These are a few of the words and phrases that ChatGPT tends to overuse.

Since its launch, ChatGPT’s writing tics and quirks have helped many create an internal AI detector.

But not all words that seem like generative AI are a sign of ChatGPT at work. Sometimes words just appear more often in other languages or dialects.

That’s why it’s important to remember this age-old rule – “Context matters.”

Here are 7 words that are frequently flagged as AI content, either as a sign of AI overuse or an AI hallucination, yet may simply be the result of how another culture speaks.

Keen

Keen is the best example of this phenomenon where internal AI detectors signal a false positive.

When we first launched the email feedback feature for AiSDR, several of our American users flagged keen as an AI word to remove from future emails. Consequently, we updated many American sales personas to include a rule that prohibited the use of keen.

At the same time, several of our European users preferred to keep the word.

Long story short, keen is commonly used in British and Australian English and a lot less in American.

Cheers

There are many ways to sign off an email, especially in non-formal English:

• Regards
• Best
• Thanks
• Thanks in advance
• Talk soon

The list goes on. You can even omit a sign-off, opting instead just to use your name.

But one sign-off you’re unlikely to encounter from Americans is Cheers. That’s because Cheers is typically reserved for toasts, which is why it’s sometimes flagged as AI when it crops up in emails.

This can easily be a false positive as Australians and Brits frequently use Cheers to say “thank you” or “goodbye”. And if you think about it, this dual meaning makes it perfect for signing off an email.

Fortnight

I was torn between the terms fortnight and score, but the former has a likelier chance of appearing in sales emails.

In case you’re unfamiliar with either term, fortnight means “14 days” while score means “set of 20”.

In the US, two weeks is the standard most people use while in the UK, fortnight is still reasonably common. However, it’s supposedly falling slowly out of use in favor of the American two weeks.

Next time you see the word fortnight appear in your email, take a look at who sent it. If it’s a Brit, it’s probably not AI. If it’s an American, well… you get the idea 🙂

Humor

Just to prevent any misunderstanding, there’s no issue with the word humor itself.

Its place on this list is more about the spelling – humor vs humour.

Although there’s a lot of content available on the internet about the many spelling, usage, and terminology differences between American and British English, this doesn’t stop people from flagging unfamiliar spellings as AI hallucinations.

Here’s a quick recap of some of the key spelling differences. 

Aside from alternative spellings, there are plenty of words that vary between the main dialects, such as chips and fries in the US to crisps and chips in the UK.

My favorite?

Barbie – A famous doll (and movie!) in the US but slang for barbecue in Australia.

Holiday

In British English, holiday is frequently used to describe vacations and other days off, not just special days of the year.

In comparison, Americans usually reserve holiday for occasions like Christmas and Thanksgiving. Instead, Americans would use terms like days off, time off, leave, or vacation.

Although it’s easily understood from the context, some flag the use of holiday as an AI hallucination when they see someone describing upcoming time off.

Crucial

For better or worse, Crucial is seen as another indicator of AI because it feels like whenever AI wants to say something is important, it uses the word crucial.

I can’t disagree. After all, Americans would use words like important, key, and critical or expressions like make or break, crux, and holds the key over crucial any day. 

However, Americans aren’t the only people who speak English. 😅

Pulling from my English teaching experience, a lot of textbooks teach non-native speakers that crucial is simply a stronger synonym for the word important. So whenever you want to say “very important”, you say crucial.

(Oddly enough, those same textbooks rarely offer the word critical as an alternative, which Americans would probably use more frequently.)

In the end, this leads many non-native speakers to rely on the word crucial whenever they want to emphasize that something is essential. And as a result, their content is flagged as AI.

Insights

Insights is a peculiar case that is equal parts buzzword, everyday word, and AI content indicator.

How?

In today’s ever-evolving world (see what I did there?), people are drowning in data and information.

With so much at peoples’ disposal, they latched onto a word that could describe specific data within data, which just so happened to be insight.

Fluctuations in the stock market from 1929 to 2020? A financial insight.

How customers make purchase decisions at grocery stores? A shopper insight. 

And how do you describe an article or blog about recent trends in marketing? Insightful.

This is where we run into the challenge of AI content. Most models are trained on a huge set of data that mirrors how people speak and write. And if a buzzword like insights is frequently used, you can be certain that AI will start using it. 

For this exact same reason, other buzzwords like gamechanger could easily have made this list.

Still, I have found that while the everyday use of insights might not be huge among Americans, it is heavily relied on in the tech sector and the international English-speaking world.

Honorable mentions

For this list, I wanted to focus primarily on words that people see and automatically assume to be signs of an AI writer.

But there were a few scenarios that I felt were too interesting to pass up, so I’ve added a small section of honorable mentions.

I hope this email finds you well

I remember in the early days of my professional career, we were always taught and told to include a sentence like this at the start of an email.

Even textbooks I used for teaching material taught the same.

Nowadays though, people who are in the business of sending emails want the text to be under 50 words and to get to the point asap. As a result, when they see an empty line like this, their first impression is “AI content”.

Some habits can be hard to break though. I still regularly see email intros like this in many incoming emails that I know aren’t AI-generated.

Oxford commas

In case you need a refresher, the Oxford comma is the comma that appears before the conjunction in a list of three or more items (e.g. bacon, lettuce, and tomato).

However, the English-speaking world is fiercely split about whether or not to use the Oxford comma (for the record, I’m in the camp that uses the Oxford comma).

By default, ChatGPT uses the Oxford comma, which has led some who don’t use it to assume that when they see it, the content must have been generated by AI.

This is why it’s important to consider the writer and where they’re from. It might even be that their organization has a specific policy for using or omitting the comma.