If you've been playing around with ChatGPT for a while now, have you noticed anything...unique...about the way it tends to write?
Maybe. Maybe not. 🤷♀️
ChatGPT, despite being a robot, does have its own "unique" writing style, due to the language patterns and structures it was trained on. (Or, that it steals from, but that's a conversation for another day.)
Though it can learn new language styles overtime, ChatGPT initially delivers certain patterns, phrases, and stylistic choices that can make its output recognizable or distinct to those who can recognize it (AKA, me and maybe your audience). But, when you are able to recognize these patterns, you can make the alternations necessary to make whatever content producing sound more like your brand or company's voice, and not like a machine.
Below is a list of what I call ChatGPT-ism including examples for each. These characteristics are not exhaustive and can vary based on the model's training data, version, and the specific prompt it responds to.
I even included some sprinkled throughout this article to see if you can catch them. 😊
Will People Know That I Used ChatGPT to Write My Content?
This is a question that people ask all the time. More often than not, we see people posting things like, "Wow, this is so obviously written by ChatGPT" and usually, it is. People who put a prompt through and publish exactly what Chat spit back at them will usually get feedback very quickly. And, that feedback isn't always the best.
But, does it matter?
To the untrained eye or for those looking to get information fast, not really. If content really isn't your game but something you just need for some of your marketing materials, then it's not the end of the world if it's very clearly written by Chat. It all depends what your goals are.
We'll get to that more later on.
However, there are instances where it very much matters.
A cover letter.
A SEO blog post you want to rank and stay in position.
A LinkedIn post.
A landing page.
You get the idea.
In these situations, it's important to proofread what Chat gives you in order to make sure your content looks and sounds as original possible. Especially with the new changes to Google, you'll want to to keep an eye out for common ChatGPT-isms to ensure you're producing high-quality content.
The Most Common ChatGPT Outputs (And, What To Write Instead)
1. Frequent Use of Qualifiers and Modifiers
AI is "very likely" (see what I did there?) to use qualifiers and modifiers to present information in a nuanced manner. Qualifiers and modifiers are both tools in language that help us add detail or adjust the meaning of sentences. Think of modifiers as the big category that includes anything that adds more information to other words, while qualifiers are a specific kind of modifier that fine-tune or change how strong a word's meaning is. So, while all qualifiers are modifiers, not all modifiers are qualifiers.
If that sounds like a bunch of gibberish, basically what it means is that qualifiers tweak the intensity of words, and modifiers add extra info or detail to make sentences clearer or more interesting.
Essentially, while modifiers are important to making the text more descriptive, the use of qualifiers to make the text more descriptive is often overdone by AI.
Some common qualifiers are:
"Very"
"Pretty"
"Really"
"Quite"
"Rather"
"Somewhat"
"Extremely"
"Slightly"
"Particularly"
"Fairly"
"Highly"
"Too"
"Enough"
"Almost"
"Just"
"Completely"
"Absolutely"
Example: "It is somewhat likely that the new policy will have a significant impact on the market."
What To Do Instead:
Qualifiers are sometimes necessary to emphasize the point you're trying to make, but other times, it's just fluff. And, because AI tends to rely on these more often than not, your text can end up being over saturated with qualifiers. So, read it back to yourself, and ask "If I remove this qualifier, does it drastically change the meaning of the text?" Chances are you can find a few spots to delete them.
2. A Neutral and Balanced Tone
AI isn't trying to start fights with anyone. So, it does this by maintaining a neutral tone, especially when discussing controversial topics. If you're trying to write a piece that's able to give a dual perspective or present two-sides of an argument, or you're looking to write a piece that's strongly opinionated, you'll find that can Chat can do it, but you may leave reading it feeling a bit like you're not convinced to go one way or another. It loves to do this in summaries/conclusions most of all.
Example: "Both sides of the argument present valid points, and it is important to consider the perspectives of all stakeholders."
What To Do Instead:
While you can use ChatGPT help lay out and present your argument(s), a lot of the points that you're using those arguments to make can be difficult to accomplish with Chat. So, use it to help you outline or organize the points, but you'll have to go in yourself to make and/or emphasize the points you want.
3. A Preference for Passive Voice
AI models often use passive voice to make statements less direct. While passive voice in itself isn't an issue, it's generally looked down upon from an SEO standpoint, which prefers articles that are more reader-friendly, and that means a more active voice overall.
An example of passive voice is "The experiment was conducted by the research team", versus "The research team conducted an experiment". Even though these sentences have the same meaning, you can see how the latter is a bit more direct and punchy.
Additionally, active voice can be important in journalism for a wide array of reasons. One issue that has come up, for instance, is how we talk about victims of sexual abuse. This is because how we frame the narrative can greatly influence reader perception. Using a passive voice, such as "The victim was assaulted," can obscure the responsibility by not directly naming the perpetrator.
This phrasing focuses on the victim's experience without attributing the action to the person who did the assaulting. On the other hand, an active voice sentence like "The person assaulted the victim" places clear responsibility on the perpetrator, making it evident who committed the act while also being careful not to victim-blame the survivor.
What To Do Instead:
Of course, the example I gave is in more specific cases. But, generally, writing in an active voice just reads better. That being said, if you're not a writing-wiz, it can be hard to just switch your sentences from passive to active, and there may be instances where the passive just makes more sense and reads better. This is something an experienced writer might be more natural at identifying.
The good news is that Yoast SEO in Wordpress can help you out by actually highlighting where you're using passive voice, so you can go in and switch it, which just involves moving the object of the sentence to be the subject. Apps like Grammarly and WordTune can help with this, too. Or, once you identify the sentence, you can ask ChatGPT to help you change it, with this prompt:
"Can you change this sentence from passive tofor active?" Go ahead, try it! The more you practice, the better you'll get at identifying these and changing them naturally on your own so you can ensure any content you have written by AI is a little more human.
4. The "Three-Items-in-a-List"
AI models frequently use series of items, separated by commas, often adhering strictly to the rule of three. (Again, see what I did there?)
Example: "The new software is fast, efficient, and user-friendly" or "Austin, Texas, is known for its vibrant culture, rich history, and numerous attractions that make it a unique and lively place to visit or live."
Seriously; it's the one thing I notice constantly that ChatGPT does and for me, it's the one pattern that's a dead-giveaway that Chat was used to write content. Again, depending on your goals, this isn't the end of the world, but taking a little extra time to address it can make your content more original.
What To Do Instead:
Look for it, spot it, and change it. If you notice that Chat is spitting out three items/examples in a sentence like that, get rid of one, combine two, add one, or add a lot more text at the end so it's not as noticeable. You could also break it into two sentences. It's also likely that Chat easily pulled this straight from another article, so it's not original. You can add your own information to it.
If you're having trouble, ask Chat to specifically NOT do that by saying, "Can you get rid of any sentences that use three items at the end of the list separated by commas?"
Look what I got now: "Austin, Texas, is known for its vibrant culture and rich history, offering numerous attractions that make it a unique and lively place to visit or live."
See, a little better! But, again, you could make this your own.
5. The Use of Technical and Precise Language
AI tends to use technical terms and precise language to convey information accurately.
Example: "The algorithm utilizes a convolution neural network to enhance image recognition accuracy."
I asked Chat why it does this, and apparently it's because the goal is to relay information as accurately and unambiguously as possible, especially on topics requiring specific knowledge or expertise. So, if you need to do some very technical writing, then Chat is your best friend. But, if you want your piece to be more conversational, authentic, and ultimately, reader-friendly, you're going to want to scan for overly technical language.
What To Do Instead:
One of the reasons that certain industries need the help of (human) content writers is because the goal is to break down complex content into digestible content that readers can actually understand. This is something that was always very important in jobs I did for cybersecurity, IT, IAM, and companies in related industries. A good rule-of-thumb? Ask another adult if they understand what the sentence means. (This is great advice for SMEs or technical editors who need to contribute to content, but have a hard time breaking out of the truly technical verbiage.)
If the answer is "no", try to break it down further. Again, you can ask Chat to help you with this part, by prompting it to say "Can you make this more reader-friendly and approachable so that it makes sense to the average reader who may not have so much background knowledge on this topic?"
6. Repetitive or Redundant Words & Phrasing
AI tends to re-use certain words and phrases over and over again. Personally, this drives me absolutely bonkers. When I see that the text has these, I know immediately that it's written by ChatGPT. At the same time, these words are not unusual to use in certain contexts, but for the mere reason that Chat uses them regularly, it makes me not want to use them when I'm writing original content. Some words/phrases I notice a lot are:
Crucial: "Understanding the basics of digital security is crucial for anyone using the internet."
Landscape: "The technological landscape has evolved rapidly over the past decade" or "In the ever-evolving landscape".
Notable: "A notable advancement in the field of medicine has been the development of personalized treatments."
It's worth noting: "It's worth noting that the study's findings need to be replicated before drawing definitive conclusions."
Moreover: "Moreover, the project will not only benefit the local community but also set a precedent for future initiatives."
Considerable (or, considerably): "The project requires a considerable amount of time and resources to complete successfully."
Bear in mind: "Bear in mind that the success of this strategy depends heavily on market conditions."
These are just a number of examples I've come across through my work. I'd love to hear any you have found! (And, that just makes you cringe when you see them.)
What To Do Instead:
If I find that ChatGPT keeps giving me repetitive words or phrases, I simply just change them. But, not all of them, all the time. Sometimes, these words or phrases simply make the most sense depending on the context. But, using them more than once is not only an indication that it was written by Chat, but it's not always great writing in general. A good rule-of-thumb is to mix it up!
7. The Emphasis on Logical Structure
Remember when you were in primary school (or, was it middle school?) when your teacher explained "sequencing" to understand the logical structure of how a story should be written?
Well, ChatGPT would get an A+ on that exam! That's because AI's responses tend to be logically structured, with a clear beginning, middle, and end.
Example: "Firstly, we will examine the historical context; secondly, we will analyze the current situation; finally, we will predict future trends."
This example is a shortened version, of course. I'd never recommend that you would include this in a content piece at all, as it's better to get right to the point. (Unless you're writing a pillar page or other long-form content, then I recommend using jump-links to the various section headers from a bulleted table-of-contents list in the intro.)
But, for the sake of this article, ChatGPT can rely on using this kind of sequencing throughout an entire piece of content, or within individual sections when highlighting points. Sometimes, it's necessary to do this, for instance, in more academic writing (though, you should be super careful using AI for that, as it can violate academic integrity) or writing a recipe or a "how-to" article. But, there are many other ways to change this up so it's not so staccato (and, obviously written by chat). Read below.
What To Do Instead:
Generally, using sequencing in articles depends on the situation. It's also important to use transitions in your article for flow, which Chat is sometimes not great at naturally (which is the point to having flow altogether). But, see if you can change up the words used. Here are some suggestions, but keep in mind that again, it depends on the context. Are you listing events in order or are you making points in order?:
First/ly can be changed to: "To start", "To begin with", "First of all", "For starters"
Second/ly can be changed to: "Next", "After this", "Afterward", "Futhermore" (or, keep "second" and change the others!)
Lastly can be changed to: "Finally", "Last but not least", "Ultimately".
As long as you mix-and-match, you'll be good!
8. The Use of Phrases Indicating Uncertainty or Probability
AI frequently uses phrases that indicate uncertainty or the probabilistic nature of its statements.
Example: "It is likely that...", "There is a possibility that...", "It may be the case that..."
What To Do Instead:
If you're trying to establish yourself as an authority in your industry, the last thing you want to do is be uncertain about the points you're making. If you're not certain, leave it out altogether. Or, if it's something scientific that's still being studied, mention this with sources.
But, if this is for more of a creative writing piece or thought-leadership piece (which, again, I'd advise against using Chat as much as you can, though it could maybe assist with organizing your ideas), then try changing those phrases to something like, "There's a good chance that...", "It may seem as though", etc. This gives it a more casual vibe that's more approachable for readers, and doesn't scream "I was written by Chat!"
9. A Tendency To Be Excessively Formal
AI's language can sometimes be more formal than necessary for the context.
Example: "It is of utmost importance to ascertain the veracity of the information presented herein."
Yes, this might be something that Chat tells you to do after asking it to write you a piece that's attempting to prove a point based on alleged facts. (Remember, Chat doesn't do well with pulling statistics, unless you give it to it straight. Even then, their can still be issues.) But, this is an example of how technical it can be, even if you're not aiming for that.
What To Do Instead:
Pretty much anything but. It's amazing; sometimes even if the piece is more lighthearted and less-academic, Chat will sneak a sentence like this in. For content writing; blogs, landing pages, email marketing, there's usually no need for this type of language. So, be sure to delete it altogether or replace it with something that fits the rest of your article's voice and tone before publishing. You can also use Hemingway to make sure your content reads at the appropriate level for your audience.
10. Adding Unnecessary Definitions or Explanations
AI often includes definitions or brief explanations of terms it uses. Though definitions are necessary at times, it's something you want to look out for.
Example: "Blockchain, a decentralized ledger technology, enables secure peer-to-peer transactions without the need for a central authority."
What To Do Instead:
It's crucial (ha) that companies who are speaking to their audience via content or copy understand the background knowledge that that audience has. You never want to assume your audience knows everything, nor do you want to unintentionally insult them by saying something that's obvious to them. If the above example is catered specifically to people who trade cryptocurrency, then it's absolutely unnecessary to define it.
General Tips to Consider When Using AI to Write Your Content
In addition to following some of the tips above, here are some general points to keep in mind when relying on ChatGPT to help you with your content marketing:
Try to write as much of the piece as you can on your own: Before you even get started, if you have any ideas of how the piece should be written, go for it. Even if you end up using ChatGPT, being able to provide it some of your original thoughts/writing can go a long way.
Don't re-prompt more than 3x: If you keep telling ChatGPT what to write and you're not getting what you want after three attempts, you may need to go back to the drawing board.
Lower your expectations: ChatGPT is good, but it's just not there yet. If you're not a strong writer or don't quite have the budget to invest in a content writer yet, keep in mind that Chat won't be perfect.
Read the text back to yourself: This editing trick is as old as time. Read the article out-loud to yourself before publishing. It's one of the best ways to catch any mistakes, or any of these common ChatGPT tendencies.
Use this guide!: If you want your AI-written content to sound more original, print out this guide and keep it as a handy reference.
Make Your AI Writing Look Original
As a writer, I will never 100% encourage the use of AI tools for writing. BUT...also as a writer, I need to keep in mind that people are going to use it regardless to help meet their content marketing goals more efficiently.
But, at the very least, do your best to put in that extra step to make your article look more original by checking for these common ChatGPT-isms. This will help your article appear more authentic, and depending on what your goals are, can still do well on search engines without getting flagged.
Need some additional support with your writing needs? Email me at hanalarock@gmail.com