#5 Humanizing AI Text for Academic Papers (No Tools Needed!)

Hi, this is Chelsi from Ref-n-write. In this video, we'll talk about how to humanize AI generated text. I want to be very clear.

We don't recommend using AI generated content in any of your academic papers. However, we are living in an era of AI, and there might be times when you want to incorporate text snippets from ChatGPT into your work. If you choose to do this, be extremely cautious.

AI detectors can flag such content and you will be accused of committing plagiarism. It's very important that you rewrite and humanize the text before including any AI generated material in your paper. Before we look into ways to humanize AI text, you should first understand the basic characteristics of AI text and how AI detectors manage to identify them.

I have asked ChatGPT to write a few sentences about social media and this is what I got. Let me copy the text and run it through an AI detector. Let's use the free version of the online tool WordPotter AI.

Let's paste the text and press the run AI check button. As you can see, the AI detector flags the text as 100% AI, which is not surprising. We did generate this text using ChatGPT.

There are four main reasons why the AI detector was able to correctly identify this as an AI generated text. Number one. Sentence length.

Number two. Repetitive phrases. Number three.

Clichés and fluff. And number four. Outdated and old fashioned language.

Let's go through each of these factors in detail one by one. Number one. Sentence length.

ChatGPT and other AI tools tend to write passages that contain sentences of similar length. If we look at the sentences in this passage, they are all between 24 and 25 words in length. This normally doesn't happen with human text.

When humans write, they use a mix of short and long sentences within a paragraph to create a natural flow. That's one of the things that make human text feel more dynamic and engaging. Number two, repetitive phrases.

ChatGPT often uses repetitive phrases in the text they generate. This makes it easy for AI detectors to identify whether the text was written by humans or AI. For example, let's ask ChatGPT to write something about mental health.

Now let's compare both pieces of text generated by ChatGPT. You will notice that there is a lot of repetition between both passages. Both passages start with the phrase “In today's world”.

Then we have the phrase “an integral part” in both passages. In the first passage, we have the phrase “before delving into”. And in the second passage, we have “before diving into”, which are quite similar.

Then we have the phrases “is essential to” and “woven itself into the fabric of daily life”, which appear in both passages. Word for word. This is one of the reasons why AI generated text is easily identified and flagged by AI detectors.

Number three, fluff and clichés. Look at the phrases “In today's world” and “Look no further”. These are what we call clichés or fluff.

It means that these phrases do not add much value to the text. If we were to remove them, the text would still read fine. If you had spent some time with ChatGPT, you might have probably noticed that ChatGPT loves to load its responses with clichés and fluff.

You must remove these phrases to humanize AI text. ChatGPT often uses vocabulary that is not typically used in everyday conversations. The word choices can sound overly formal and outdated, something you would typically find in old books or official documents.

For example, take the phrase “Before delving into the benefits of” we don't say this in normal conversation. We would probably say “Before we get into the benefits of”. Instead of saying “it is essential to grasp the origins of”, we'd say something simpler like “it's important to understand where it all started.

” Look at this sentence ChatGPT has generated. Social media is here to stay and conquer it's reach across the globe and has woven itself into the fabric of daily life. Who talks like this?

It sounds formal and poetic at the same time. We could simply say Social media is now everywhere and is part of our everyday life. Here is the rewritten passage.

As you can see now, it sounds more direct and natural. The sentences vary in length and we've cut out all the clichés and unnecessary fluff. Now let's put this modified version through the AI detector and check the results.

The detector classifies the text as human written, which makes sense because we took the original ChatGPT text and made substantial changes to humanize it. And it worked. If you have any questions, please drop a message below.

We will answer it as soon as possible. If you still haven't installed Ref-n-write software, please do it. It comes with referencing tools, plagiarism checkers, proofreading tools paraphrasing tools and academic phrase bank, and many other useful features to help you write your research papers and theses.

You can sign up for free using the link below in the description. If you are interested in training courses on research and academic writing, please visit our learning academy on Udemy You can find the link in the description below. Finally, please subscribe to our channel to get future updates.