Tech | Visa | Scholarship/School | Info Place

How to tell if something was written by ChatGPT

As we all accept a world where AI is now mainstream and no longer a fantasy only seen in movies, the need to understand what elements of daily life and experience AI impacts has never been greater. bigger.

OpenAI’s ChatGPT is the first mainstream example of how useful AI can be to a variety of people, businesses, and industries, but that doesn’t mean it’s perfect, accurate, or the ultimate assistant for our daily needs.

ChatGPT has also been somewhat controversial because it occasionally gets people into trouble by providing incorrect, inaccurate, or unfactual answers to questions. It even caused trouble for its creator, OpenAI.

In 2023, it falsely claimed that an American law professor made inappropriate remarks and sexual advances toward a student during a school trip, and even cited the Washington Post as the source of the information. There were no school trips and no articles in the Washington Post. In short, it’s complete nonsense.

This is just one of several very concerning flaws in the chatbot’s impressive capabilities. Overall, it can be a very useful source of information, but what it says cannot and should not always be taken as fact.

But as more and more people use ChatGPT to complete daily tasks online or as part of their job, there’s a good chance that every time you go online, you consume at least one piece of information generated or written by artificial intelligence, and most likely ChatGPT itself.

As a result, the internet is quickly becoming a minefield where the lines between humans and AI-written text are blurring. At some point in the future, this may no longer be an issue – ChatGPT-5 may be coming – but for now, people still need to be vigilant and know when something has been written by ChatGPT.

What are the signs that something was written by ChatGPT?

Because humans prompt ChatGPT, the details of the chatbot’s response often depend on the details of the prompt it gives. Without guidance, especially in more detailed topics, it is more likely to return vague or inaccurate responses.

This may not be obvious to the untrained eye, like someone without specific background knowledge on the subject. But to others, it was obvious that the text was written by ChatGPT.

These are the key things to note…

General use and repetition of language

ChatGPT is what’s known as “narrow” AI, meaning it can’t understand or replicate human emotions or behavior, and it can’t think for itself, meaning its responses often lack any personality or creative language.

Furthermore, even if it does make mistakes, it is trained to make as few of them as possible. Therefore, responding in a simple and somewhat robotic tone is intended to help minimize errors and inaccuracies.

This is obvious if you ask it to write a review of your favorite movie or product, for example. It may leave out critical information about the content being reviewed, such as the name of an actor or the dimensions of a product.

So if you read something similar on a review site and it seems to omit what is typically considered key information, there’s a chance it was written by ChatGPT.

The same applies to the repetition of words or phrases. ChatGPT is trained on large amounts of data and language, but it may still use repeated language, especially in larger blocks of text.

hallucination

The above case involving a law professor is a perfect example of an AI hallucinating, completely making something up in its reactions. It remains one of the biggest problems with generative chatbots like ChatGPT.

AI experts recommend fact-checking ChatGPT responses, especially when it comes to more niche information and topics. There are many other cases of hallucinations, some more serious than others, so it’s best to double-check your responses with other sources of information before trusting their accuracy.

It’s easier to spot this if you’re reading about a familiar topic. For example, a match report from a recent football match you watched might easily allow you to identify any factual inaccuracies. But if you’re studying the thermic effect of food, it can be difficult to tell whether it’s human-generated or AI-generated.

copy paste error

This is arguably the easiest to spot. People have been known to accidentally copy and paste ChatGPT replies and their accompanying comments, such as “Of course, this is a movie review of…”.

This is a method of determining whether what you are reading was written by ChatGPT or a human, and is based purely on human error, not errors on behalf of artificial intelligence.

Read the text carefully

ChatGPT is trained to sound and react like a human, so if you only read a sentence or two, it’s nearly impossible to tell whether something was written by an AI. It will become clearer if you read the entire article carefully, which may highlight specific hallucinations, repetitions, common language, or copy-paste errors.

Of course, humans can edit ChatGPT’s responses to remove much of the above and make the text more human-like, but this will usually involve a level of editing that may mean it’s easier to write the text yourself.

How to detect ChatGPT content

The rise of ChatGPT and similar AI chatbots has led to a series of AI content detectors that all claim to be able to detect whether a piece of text was written by a human or an AI.

We’ve looked at the best AI content detectors that can even indicate exactly which parts of text are human and which are AI. Some will give you an estimate of the percentage of humans vs. AI in the text for analysis.

However, none of these AI content detectors are 100% perfect, and there will always be cases where human writing is incorrectly detected as AI, and vice versa.

However, using an AI detector is a great way to see if there are signs of AI in what you’re reading, especially if you’re still doubtful after checking for repetition, hallucinations, and general language.

But the key point, especially when it comes to niche topics you’re not entirely familiar with, is to always fact-check what ChatGPT tells you. Better safe than sorry.

Featured Image: Ideographies

#written #ChatGPT

Leave a Reply

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

Index