The Sentinel

The Student News Site of Monsignor McClancy Memorial High School

The Sentinel

The Sentinel

How to Recognize ChatGPT Writing

How to Recognize ChatGPT Writing

ChatGPT is the world’s most well known AI model for many reasons. One is that people use it in school all the time. Another is that it’s also famous for confidently stating false information. This raises the question: how can you tell what’s been written or created by an AI, and what’s real?

Since it was originally released back in the long-forgotten year of 2022, AI technologies have been immensely popular throughout the world and especially on the internet. During the earliest days of its existence, a lot of interactions with it would be just people making fun of its mistakes. At the same time, they would be very surprised at what it was capable of. Mostly, it would continuously repeat that it was “an AI model trained by OpenAI” and could not perform certain things. But by 2024 it’s progressed significantly, hallucinating (making inaccurate statements) less and seeming to make more sense overall.

While it has been shown that there are benefits to AI like writing code faster in computer programming, there is a lot of evidence that proves people use it to cheat in school. For example, the usage of ChatGPT went down a lot during the summer after it was released. People originally thought that this was because the original excitement from the technology had died down, but when usage of it increased again in late 2023, it was clear that many students were using it in school. So how can you distinguish between what was written by a real person and what was created by AI?

For the most part, there’s no foolproof way to recognize what is written by an AI like ChatGPT. There are lots of things that can be written that are simple or unremarkable enough that it’s difficult to notice, and most software that is used to detect it is imperfect. However, some common things that you can find AI saying are things like having suspiciously complicated vocabulary, hallucinating information, repeating the same words throughout a paragraph, and contradicting itself in some places. OpenAI themselves are also going to launch an AI classifier sometime in the near future to help identify what is written by AI and what was written by a real person.

 

Sources:
How to Detect OpenAI’s ChatGPT Output | by Sung Kim | Geek Culture | Medium

Mozilla Foundation – Did ChatGPT Write This? Here’s How To Tell.

ChatGPT – Wikipedia

Leave a Comment
More to Discover
About the Contributor
David Rojas, Columnist/Editor
Hello. I'm David, and I'm not an official journalist yet, but until then I like learning about things that interest me, explaining those things to others, and always searching for new & exciting experiences.

Comments (0)

All The Sentinel Picks Reader Picks Sort: Newest

Your email address will not be published. Required fields are marked *