ChatGPT, version 4o or else, can write very elegantly under most criteria. It produces immaculate, grammatically correct writings… which could be gibberish. Since it produces paragraphs (and bullet points) that look so correct, many people are tempted to consider its write-ups useful. They’re usually far from that.

That poses an additional problem for programmers. Developers who ask ChatGPT for help will never get an “I’m sorry—I’d ask a colleague of yours about this.” Rather than that, any chatbot—not just ChatGPT—will go off in a writing frenzy and turn out code. That doesn’t mean that the code, that looks pristine, will work. Studies are showing that most of the ChatGPT answers for developers are wrong, and almost 8 out of 10 are verbose. They look grandiose. But they’re fools’ gold.

Still, that’s something to worry about once you get the job. If you’re a developer seeking employment, you can always rely on LLMs for your cover letter. But you don’t want your job application to give away an effortless, buzzword-laden, even incorrect statement. We advise that you seek out the signs to detect if something was written by ChatGPT and do something about it.

Is ChatGPT Getting Worse Over Time? -->

How to detect ChatGPT writing    

Most writers can easily tell when something is written by AI, especially when the user is lazy enough to simply input, “Write me an article about XYZ” or “I need an Instagram caption about XYZ”. However, if you’re someone who doesn’t have a background in writing or editing, you probably ask yourself fairly regularly, did ChatGPT write this? Or, is there a way to detect ChatGPT written content? And the answer is yes, yes, there is.

For the sake of this article, I’ve asked ChatGPT-4o to write some text. We then break down the text to provide examples of what to look for.

First, we’ll go through the typical signs of most Large Language Models. This is how you detect ChatGPT write-ups:

• Bullet points all around, even when they don’t serve the purpose: LLMs seem to prefer bullets to standard paragraphs. This got out of hand when Google’s Gemini dropped. ChatGPT 3.5, for example, wasn’t that prone to writing bullet points for everything. This whole section is written in bullets, so that’s for you to consider: Was it actually written by a machine? You can easily tell.
  ‍
• Hallucinations and factual inaccuracies: It’s a challenging time to surf the web, because it’s hard to spot a factual inaccuracy when you don’t know about the topic in the first place. But LLMs are now telling you to eat pizza with non-toxic glue. When you come across such a statement, you can tell it’s an AI because it goes against common sense. Large Language Models don’t have a symbolic understanding of how the world works (i.e. their images don’t usually agree with physics), and so they can’t reason that their advice can hurt someone. They’re just very expensive and energy-consuming parrots.
  ‍
• A knack for complicated words: Delve, cutting-edge, rich tapestry, “ever-evolving landscape” and explore will show up plenty of times, even if it doesn’t fit the situation. You don’t go around calling C++ “cutting-edge software.” Large Language Models will prefer to elaborate on something “tailored for” or “catering to” even if that verbosity was uncalled-for. To be fair, plenty of English speakers in Nigeria use “delve” daily. But if you’re not from Nigeria or are not interacting with a Nigerian programmer, look out for it. There’s nothing wrong with it per se, but it has become an indicative sign of a low-effort, automatic write-up.
  ‍
• American spelling: Many Europeans write their correspondence in British English, but ChatGPT might make cover letters in American English with a lot of ‘z’ characters and very few ‘u’ characters. This is probably because OpenAI “fed” their engine with easy-to-access content in US English, like most of Reddit or The New York Times is written in.
  ‍
• Disclaimers that show up frequently: People have said that they’ve been reading academic papers that start with the phrase: “As a large-language model trained by OpenAI, I can’t…” or even the phrase: “Regenerate Response.” That’s how pervasive this clue is. It’s the disclaimer OpenAI gives you when you ask for something that the solution can’t do.
  ‍
• Surplus brackets: Chatbots and Pete Doherty have something in common: They both brought brackets back from oblivion. If you ask it to write a template, it will have brackets all around. If someone forgets to remove or fill in these brackets, they’ll be giving away a no-effort ChatGPT write-up.
  ‍
• Wordiness, because time is not of the essence: Chatbots aren’t held accountable if they go over word-count limits or if they’re verbose because they don’t invest time as we do. They’re usually long-winded. Text can be incredibly long and vague, and against the notion of being specific.
  ‍
• A remarkable absence of meaning: Large Language Models struggle to say something helpful. They cannon-fodder words and fight long and hard to say less than the Lorem Ipsum boilerplate text. Usually, they’ll give advice such as “highlight your skills.” That’s not real-world actionable guidance.

‍

Example 1 — How to get a job as a programmer

We asked a very expensive chatbot (we won’t tell you which one—but let us clarify they all suffer from the same problems) to tell us how to get a job as a programmer. Check the machine’s answer:

‍

As expected, it abuses from bullet points, it says vapid words, and gives directions to “tailor.” To be fair, these bullet points are not that particularly bad. They are good pointers as to where to start. But they’re superficially unhelpful. If you were to ask the machine to expand on each, you’d get tonnes of gibberish that looks very neat. Building an online presence takes heaps of time, but since the LLMs can’t be held accountable, they’ll hurl the advice away as if it were a walk in the park.

When you’re using ChatGPT to your greatest advantage, use it up to this point, and then just leave it behind. You now have an idea of how to proceed. Now keep the decision-making human. They’re better writers, but their understanding of how you interact with the world is terrible.

Find JavaScript Developer Jobs →  

Example 2 — Introduction paragraph 

We asked ChatGPT to “write me an introduction paragraph for an article called ‘How to tell if something was written by ChatGPT’.”

‍

As we said earlier, ChatGPT writes elegantly. However, the above text features overly complex sentence structures such as, “Understanding the telltale signs that reveal its hand in crafting a composition becomes both an art and a science.” 

And an unusually formal tone, “We delve deep into the intricacies of language and the subtle nuances that might betray the digital origins of a piece of text”. Also, “Join us on this exploration”, you’d think we were taking a trip to Antarctica, not reading about how to detect ChatGPT writing. “Fascinating challenge”, we wouldn’t say this topic is fascinating, would you? If what you’re reading features unusual or incorrect wording, that’s also usually a red flag. 🚩

Example 3 — Health article

We asked ChatGPT to “write me an article about Intermittent Fasting”. Firstly, the article it produced was nearly 3,000 words long, which is another telltale sign. AI-generated articles are usually unnecessarily long and wordy. They also don’t always make sense. For example, this is a section from the article:

‍

Benefits and Mechanism? Really, ChatGPT? Unless you’re writing a Thesis, no one talks like this. And to add, if you’re wondering how to detect ChatGPT writing in a Thesis, the lack of credible sources should be a big hint. 

“This eating plan works by prolonging the period when your body has burned through the calories consumed during your last meal and begins burning fat.” Again, this sentence is quite complex and won’t make sense to a general audience. Rule of thumb: if it’s complicated to read and seems to ramble with intricate sentences that don’t really make sense, it’s probably been written by ChatGPT. 

“In summary, intermittent fasting has numerous potential health benefits including longevity promotion, cognitive enhancement, weight management assistance, and disease prevention/reversal.” ChatGPT's summary simply repeated the “benefits and mechanisms” listed above without actually summarising the section. So watch out for repetitive words or phrases, as these will likely suggest AI-written content.

Find English-Speaking Jobs in Europe →

Example 4 — Promotional email 

We asked ChatGPT to “Write a paragraph for an email promoting a 50% off storewide sale on gluten-free products.” Here it goes:

‍

We think this one’s quite obvious. The text is consistent and fluent but lacks the personal touch of naturally written text. For example, “Dear Health Enthusiast”, we don’t know any company that would address their customers like this. 

“Our gluten-free health products” and “From wholesome gluten-free snacks to nourishing supplements, we've got everything you need to support your healthy lifestyle” are vague statements that don’t provide much information. The text reads like a cheesy TV infomercial that’s impersonal and downright cringe-worthy.

“Indulge in guilt-free treats, discover new flavours, and explore a wide array of natural, gluten-free options without breaking the bank. It's our way of making your health journey more affordable and enjoyable. Hurry, stock up on your favourites, and embrace a healthier you with a shopping experience that's light on your wallet, but rich in benefits.”  These three sentences are redundant (repeat the same information). This usually happens when ChatGPT isn’t given specific instructions, so it’s using “fillers” to bulk out the text.  

‍

Example 5 — A more specific input 

For the first few examples, we kept the brief simple and vague. Even so, as you may know, ChatGPT’s output is only as good as the input. In other words, give the chatbot more in-depth details, and it should produce better quality content. Let’s give it a go. 

WeAreDevelopers is Europe’s number 1 developer community, where you can browse developer jobs from companies across Europe. They currently have 2,400+ job opportunities. With a free profile, you can:  

- Access career opportunities, tech talks, and events
- Get matched with job opportunities on your terms
- Gain instant, on-demand access to thousands of hours of tech content
- Get exclusive event invites and much more

I need you to write the text for their home page. Approx 500 words. When presenting the information,  use headlines and break text into sections, rather than using a big block of text.

Use a conversational but professional tone. Avoid complexity without talking down to the reader.

This is what it wrote:

‍

There are a lot of redundant and complex sentences in the output. Also, “Dive in now and let your developer dreams take flight.” Perhaps this sentence would make sense if WeAreDevelopers were an aviation school.

For some reason, nouns like journey, embark, and realm are favoured by AI chatbots. So unless you’re reading about a group of diverse beings uniting to destroy a powerful ring, you should assume the content is AI-written if you see these words.

Example 6 – Posts for social media

Social media is a constant whirlwind—so much content, but how do you know what’s real? How can you tell if someone used ChatGPT or actually put thought into what they wanted to say? Well, we decided to put it to the test and asked ChatGPT to write a LinkedIn post promoting this very article. Here’s what it came up with:

So, what’s our verdict? Well, the emojis are a dead giveaway. Who actually uses emojis that way? And then, we’ve got the usual suspects: “with the rise of” and “stay ahead in the age of”—classic ChatGPT fillers. 

Here’s the thing: on social media, simple is better. No one’s scrolling through LinkedIn hoping to read a mini-essay. Short, punchy lines keep it real, and you can almost always spot an AI post by that structured intro with a title. Sure, some people write like this, especially on LinkedIn, but this kind of setup is a total ChatGPT tell.

Is there a ChatGPT detector?

If you’re still unsure how to tell if someone used ChatGPT writing, there are AI detectors on the market which claim to discern whether your text is human or AI-generated - AI detecting AI, oh, the irony. 

However, these AI detectors have been known to provide false positives. For example, one software claimed that The Constitution of the United States was written by AI.  

Most AI-detecting software also only claims to be 98% accurate. If that’s true, that still means one in 50 inputs may still be flagged as being AI-written, even if it’s not. 

So, to further prove the inaccuracy of AI detectors, we tested one by copying the first sentence of this article. The results: (*facepalm)

How to tell if something was written by ChatGPT

To tell if someone used ChatGPT writing, look for the telltale signs, such as:

• Hallucinations and made-up facts
• Unusually formal tone in text that’s meant to be conversational or casual
• Overly complex sentence structures
• Unusual or incorrect wording
• Text that’s unnecessarily long and wordy 
• Lack of credible sources
• Repetitive words or phrases
• Redundant sentences
• Lack of personal touch 
• Vague statements that don’t provide much information
• Nouns like journey, embark, and realm

And remember, relying solely on AI detectors could lead to wrongly accusing people, so be cautious and use more than one way to verify before taking any action. Also, you should really think if you need to call out the use of chatbots. Because it’s not necessarily a problem to use them.

Get a job as a developer with European companies

One last try. We asked ChatGPT to give us an outro for this article. This is what it cranked out:

‍

Can you see it? “Committed to nurturing, exciting journey, and insightful events” are all empty phrasings that could’ve looked impressive in December 2022, but chatbots have made sure we all roll our eyes upon seeing them. It does have incredible capabilities to write it quickly, and it wrote WeAreDevelopers’ brand name with correct capitalisation without our assistance. As a piece of tech, LLMs are impressive, but they won’t get you too far if you want to stand out.

The best way to stand out is to be able to show off your skills in places that will know how to appreciate them. If you’re looking for a job as a developer, you can always rely on our platform. We connect the best developer talent in Europe with the best European companies. Our community continuously grows and we’re offering jobs in Germany, Austria, Switzerland, the UK, and elsewhere. Head to our job board and try your luck—maybe you won’t need to even use an LLM. Good luck!