“Determining whether a piece of text was authored by ChatGPT can be a fascinating yet challenging task. As artificial intelligence continues to advance, so too does the sophistication of language models like ChatGPT. Certain hallmarks may reveal the involvement of an AI, such as impeccably coherent responses, a lack of personal experiences or emotions, and an ability to seamlessly navigate diverse topics. However, discerning the origin of a text requires a nuanced approach, considering that advanced AI systems are designed to emulate human-like language patterns. This exploration into the telltale signs of ChatGPT’s hand in writing invites a deeper understanding of the capabilities and limitations of language models in our increasingly AI-infused communication landscape.”
Wait. Stop here. Does something seem off?
That’s correct – the introductory paragraph was written by ChatGPT itself. The truth is, the more
people begin using bots to aid them in their everyday lives, the more flooded online content
becomes. We now see programmers writing code with it, copywriters using it to promote
products, students using it to write their thesis. And as project writers and creatives, our job
becomes increasingly harder – we’re now battling to preserve the human aspect and make it
recognizable. Now, being human is more important than ever.
Now, don’t get me wrong – I love using AI – it’s helpful, saves time and gives me loads of
information. But when it comes to describing ideas, understanding concepts and providing
context, it sucks. As of now, anyway.
There are, quite ironically, AI detectors which might help you decode whether a piece of writing
was written by AI or human. Sadly, they’re not that advanced and often give wrong answers.
When it comes to project writing, it’s extremely important to be as precise as possible – the
context cannot just go missing. On the other hand, if the writing requires a bit more creativity,
ChatGPT simply won’t give you the concise story you might be looking for – it just isn’t human.
With that said, let’s look into a few recognizable patterns that might help us distinguish whether
a piece was written by ChatGPT or a human.
Let’s begin by dissecting the very introduction of this article.
- Unnecessarily complicated and out-of-place wording.
“Determining whether a piece of text was authored by ChatGPT can be a fascinating yet
challenging task. “
Now, is it really that fascinating? To me, it seems like ChatGPT was either bragging about itself,
or simply incorrectly placing adjectives where they just don’t fit. Beggining sentences with an
“ing” verb also is a dead giveaway – nobody speaks like that in real life. Also, when asked to
provide proof, ChatGPT may often starts with words such as “firstly, second, moreover”, as if it’s
writing a college essay by the book.
- National geographic tone.
“Certain hallmarks may reveal the involvement of an AI, such as impeccably coherent responses,
a lack of personal experiences or emotions, and an ability to seamlessly navigate diverse topics.”
You might often notice that ChatGPT speaks from an observer point of view, as if it’s not directly
involved in the process, a bystander simply watching from a distance. Quite similar to how a
National Geographic research might speak about the animal kingdom and its “certain
hallmarks”. Its tone is too formal, the sentences are often unnecessarily long and contain
complicated wording. - Buzzwords
“However, discerning the origin of a text requires a nuanced approach, considering that
advanced AI systems are designed to emulate human-like language patterns. “
For some reason, ChatGPT seems to love certain words. “Buzzing” being one of them, others
include phrases such as “embark on a journey, tapestry, unveiling, delve into, intricasies,
exemplify, traverse” and so on. And while it might be just a pretenteous copywriter trying to
make their writing sound like a TV commercial from the 90s, it might as well be a robot in
disguise. - Doesn’t really say much
“This exploration into the telltale signs of ChatGPT’s hand in writing invites a deeper
understanding of the capabilities and limitations of language models in our increasingly AI-
infused communication landscape.”
“Hand in writing” is already a dead giveaway to me – it sounds incredibly unnatural. “AI-infused”
might sound like a great wording idea, but it just simply doesn’t give any information. Ask
yourself – what does this sentence actually mean? Is it just there to fill the space? Does it
provide more context? Does it sound human? - Capitalised Letters. ChatGPT loves titles. It also loves capitalising Each First Letter Of The Next Word Because It Just Looks So Much Better. And while it might make your project application look pretty, at some point it gets incredibly annoying.
In a nutshell:
The few dead giveaways to recognize whether ChatGPT it the author:
Toni Lyubenova
- Long text
- Lack of context and out of place wording
- Speaks as if it’s observing from the distance
- Specific words such as “embark, tapestry and intricasies”
- Too many words, too little meaning
- Capitalise The First Letter of Each Word in Titles