With the launch of text-writing AI models such as ChatGPT, creating papers, essays, business plans, and projects has never been easier.
Just a few clicks, and you have a piece that looks professional and well-composed.
However, AI-written content is facing scrutiny via AI detection tools, these tools, while imperfect, are often relied on for establishing if a piece is genuinely human-written or machine-generated.
After using ChatGPT for over half a year I can almost immediately tell when someone has written something completely using ChatGPT. It just sounds so robotic and embarrassing.
I'm not saying ChatGPT isn't good. It's actually too good. It writes so much better than a normal human does (unless you prompt it in weird/unique ways.)
So to generally "remove" the AI from your writing, you literally just have to make it look more human-like. How do you do this? Changing vocabulary, sentence structure, etc.
While I really do appreciate all the detectors and use them to vet my own writers that I hire, I've completely recognized they could be faulty no matter who you are. Both false negatives and false positives. Sometimes you're just using ChatGPT to help you get over some writers block and need to remove the "AI" writing. That's where these tools will help.
Now I'm not going to claim these are perfect, they are not. Anybody telling you such would be lying. These tools generally help to make your ChatGPT writing more human-like, but every software does have its faults. Here's how to remove AI from your ChatGPT writing:
Remove AI Detectable Writing With These Incredible Tools
AI detection tools can be a nuisance, particularly for those who don't rely on AI for writing, yet get flagged mistakenly.
Anomalies also exist where obvious AI-written content goes undetected. This issue extends beyond the academic space into various other industries where content may need immunity against detection.
The simplest measure against getting flagged by an AI detector is scrambling the words in your text. AI detection tools work by recreating your text using AI models – more variability in your text suggests a low probability of machine involvement.
However, this could all change soon as AI continues to evolve, potentially outgrowing the capabilities of current detection tools.
1) Use a Tool Like Undetectable.ai
Undetectable AI stands as a reliable tool in scrambling and paraphrasing human content. It shuffles text quite well, though it may miss the mark slightly with more sophisticated writing (or just rewrite something that doesn't make sense at all. However, it's quick, affordable, and easy.
Running the tool involves pasting your text and hitting the humanize button. The system processes your request in seconds, and you can check for any detection flags across various detecting platforms.
On a humorous note, expecting an AI model to create content that another AI can't detect is a funky paradox.
Though it may sound counterproductive, the truth remains that any output from an AI like ChatGPT is based on predictable patterns, and AI Detection tools are designed to catch them.
2) Also Try HideMyAI
Another tool like Undetectable is HideMyAI. It works almost identically to Undetectable but offers a different engine to rewrite AI writing. You can modify 550 words for free when you register for an account. I've had good levels of success switching between this and UD on different types of writing.
It clearly removes traces of AI both in the readability sense and the detector sense. You'll generally pass a lot more AI detectors and still keep the original meaning of your writing (something these paraphrasing tools can often remove).
You can modify the type of writing (like what vibe you're aiming for), the academic/educational level of reading, and perspective (like I did below). These also pass most of the leading AI detectors.
3) Paraphrasing with QuillBot
I really enjoy using QuillBot to paraphrase articles and essays, but I really only take certain paragraphs with a grain of salt.
QuillBot shares Undetectable's functionality but with an added feature. QuillBot lets you regenerate sentence structures continuously until they meet your standards.
Although it requires more tinkering, QuillBot provides more flexibility to users, allowing customization in a simple format. Some people actually like adjusting the output within the tool even better.
4) WordAI
August 19 2023 Update: I'd actually advise against using WordAI! Users have reported them to not reply to support or issue refunds on time. The tool works well but company behind it is not to be trusted (in my honest opinion)
WordAI is quite similar to Undetectable, it just has a different engine that rewrites everything under the surface. You could try them both and see how your writing looks. Here's the normal interface for a rewrite & the sample based on what I pasted into Undetectable & QuillBot:
I think both work very well. Also keep in mind that this really depends on what detector you're trying to test on.
PS: The more content you scramble, the less the sentence might make sense, but the higher chance you'll get rid of the stain of AI detection. This works the other way around too!
5) Manually Scrambling Words
While AI-produced vocabulary may come across as fine, it still possesses a predictable pattern that human writers could exploit to their advantage.
As silly as it sounds, try to literally rewrite anything you get from ChatGPT. If you're trying to use it as inspiration, have a tab open showing it and then take the content and rewrite it as your own. Don't copy, but use as inspiration.
This is pretty much the most "sure" way of getting rid of AI detection in a piece of writing, it just takes the most time because you're applying human intuition to all of this.
So Now What?
With these tools, it is possible to evade detection by these less-than-perfect detectors, but at the end of the day only human writing can perfectly evade them.
We're in the middle of a really weird time with AI generation & detection. I can't imagine how this issue is going to unfold in the school system, but it's going to be a disruptive mess.
What is your take on AI detection issues and detectors? Have you had any of your content wrongly flagged?
