When you buy through affiliate links in our content, we may earn a commission at no extra cost to you. Learn 
Was this written by AI?
With AI tools like ChatGPT getting better at mimicking human writing, it’s becoming tougher to tell if something was written by a person or a machine. This brings up some important questions about how much we can trust what we read online.
There’s no way to 100% accurately figure out if something is produced by AI or not, but there are some tips you should keep in mind. Here’s how to detect ChatGPT text – or at least try.
Key Takeaways
- AI-generated text often shows signs like repetitive phrasing, overly polished language, and a lack of personal anecdotes.
- Tools like GPTZero and CopyLeaks can help detect AI-generated content but aren’t very accurate.
- As AI models improve, distinguishing between human and AI-generated text is becoming more challenging.
- Detection tools should be used alongside manual analysis for the best results, but they aren’t a great way to check for ChatGPT usage.
- There is no built-in watermark in ChatGPT to indicate that it was used.
What is ChatGPT?
ChatGPT is an AI tool created by OpenAI that’s designed to generate text that (is supposed to) sound like it was written by a person. You can think of it as a really advanced chatbot that can help with things like answering questions, drafting content, or even just chatting.
It’s trained on a massive amount of text from the Internet, so it knows a little bit about almost everything. You’ll find it being used in a lot of different ways, like helping businesses with customer support to generating articles.
But at the end of the day, it’s still just a program following patterns – it doesn’t actually understand the words it’s using like a human would.
How to Tell If Something Was Written by ChatGPT
Figuring out if something was written by AI can sometimes be tricky. AI-generated text often comes across as super polished, almost too perfect. You might notice it uses the same phrases or sentence structures over and over, which can make the writing feel a bit repetitive.
AI also tends to stick to general information without diving into any real personal stories or unique insights. The tone can be a giveaway, too – it might switch between being oddly formal and then suddenly casual.
This isn’t a foolproof way to tell if something was written by ChatGPT or not, but you can start to get a sense of when something might have been created by AI.
6 Signs That Something Was Written by ChatGPT
These signs don’t necessarily mean the content was written by AI, but they may give you a clue.
Repetitive Phrasing and Syntax
If you notice the same phrases or sentence patterns popping up again and again, that’s a clue. AI tends to reuse the same structures, so the writing can feel a bit repetitive or robotic.Overly Polished or Formal Language
Sometimes, AI-generated text can sound almost too perfect – like it’s been edited to be super formal, even in situations where a more casual tone would be normal. If the language feels stiff or overly polished, it might be AI at work.Lack of Personal Anecdotes
AI doesn’t have personal stories or unique viewpoints, so its writing often feels generic. If you find the text lacking in personal touches or deeper insights, it’s probably because it was generated by AI, which can only mimic, not actually relate to, human experiences.Inconsistent Tone and Style
Human writers generally have a consistent tone, but AI can sometimes flip between being formal and casual or just not stick to one style throughout the piece. This kind of inconsistency can make the writing feel a bit off.Generalized Information
AI is good at covering the basics, but it often doesn’t go very deep. If the content feels like it’s only skimming the surface without getting into the nitty-gritty details, it might be AI-generated.The writing might feel a bit shallow, missing the depth you’d expect from someone who really knows the subject.
Unusual Punctuation and Formatting
AI text sometimes (but pretty rarely) has weird punctuation or odd formatting – like misplaced commas or strange spacing. These little quirks are sometimes ingrained in us writers, as well, so this is a very minor thing to look out for.
Methods for Detecting AI-Generated Content
Manual Analysis
When you’re reviewing text to figure out if it was generated by AI, stick to the signs and patterns that we covered earlier in this article. Focus on the key things like consistency, depth, and tone.
Just use the guidelines you know, and look closely for anything that stands out as machine-made.
Automated Detection Tools
There are several tools out there that help detect AI-generated content, but they’re almost all not going to be very accurate.
These tools work by analyzing text for patterns that are common in AI writing, like repetitive phrases or oddly structured sentences. They use algorithms trained to spot these patterns by comparing the text to examples of both AI-generated and human-written content.
They usually give you a score or highlight sections that seem suspicious. This helps you figure out if something might have been written by an AI. Again, they’re not perfect, so it’s a good idea to combine them with manual checks.
After all, AI detection tools thought that the U.S. Constitution was written by AI.
Is There a ChatGPT Detector?
Yes, there are ChatGPT detection tools out there that try to detect if something was written by ChatGPT or a similar AI. These tools look at things like word choice and sentence structure to figure out if the text might be AI-generated. However, as we’ve mentioned several times, they’re not always spot-on.
Since AI models like ChatGPT keep getting better, it can be tough for these detectors to stay accurate. Sometimes they might flag human-written content as AI, or they might miss AI-generated text altogether.
It’s best to use these tools alongside manual checks to get a clearer picture. Any ChatGPT scanning tool won’t be very accurate.
Tools to Check If an Article Was Written By ChatGPT
There are a few free ChatGPT detector tools out there that claim to help you check if ChatGPT was used to generate the text.
Here’s a quick look at some of the most commonly used ChatGPT output detectors:
An important note: OpenAI, the company behind ChatGPT, had its own AI Text Classifier. This tool was designed to spot text that was written by their models.
However, the tool was only up for 6 months. They closed due to its low rate of accuracy. Here’s a quote from OpenAi.
“As of July 20, 2023, the AI classifier is no longer available due to its low rate of accuracy. We are working to incorporate feedback and are currently researching more effective provenance techniques for text, and have made a commitment to develop and deploy mechanisms that enable users to understand if audio or visual content is AI-generated.“
This should be taken seriously when you’re using an AI detection tool. If the folks who created ChatGPT can’t accurately create a ChatGPT checker, then how accurate are these other tools? My guess would be not very accurate at all. You can’t really trust software to detect ChatGPT… at least not right now. And any website that claims to detect ChatGPT with 100% accuracy isn’t honest.
Limitations of Detecting AI-Generated Text
Detecting ChatGPT text ain’t easy! One big challenge is that AI models like ChatGPT keep getting better at sounding human, which makes it harder to tell if something was written by a machine or a person.
The current tools aren’t very good either, as discussed earlier. They’ll mistakenly flag human-written text as AI-generated, or they’ll just miss AI-generated text completely – especially when the text is shorter and they have less to go on.
Hopefully, they’ll catch up with plagiarism checkers, which are very good at their job (but face a much simpler problem). We’re still a long way from having an automated tool that can test for ChatGPT-generated content.
The Bottom Line
It’s becoming tougher to tell if something was written by a human or not. Knowing how to identify AI-generated content is important for making sure we’re aware of where our information is coming from.
Roll up your sleeves and get ready to get dirty, though. AI detection tools have a long way to go.
Did ChatGPT write this article? In “today’s digital landscape, it’s critical to understand,” but we may never know (but the answer is a firm “no”).
FAQs
How to check if the text is written by ChatGPT?
Is there a tool to tell if ChatGPT was used?
Can ChatGPT be detected?
Can people tell if you use ChatGPT?
Does ChatGPT have a watermark?
References
- AI wrote the US Constitution, says AI content detector | by Michelle Harwood | Medium (Medium)
- My essay being flagged as AI when it is 100% human written (Reddit)
- AI writing – Turnitin Guides (Guides.turnitin)
- AI Detector – the Original AI Checker for ChatGPT & More (Gptzero)
- AI Detector Tool | AI Checker for ChatGPT, Gemini & Claude (Copyleaks)
- OpenAI’s AI Text Classifier no longer available due to ‘low rate of accuracy’ (Searchengineland)
