Is it possible to make ChatGPT undetectable from AI Detection? The best way to make ChatGPT undetectable is to only use the AI tool to help you with ideas. If you write the text yourself, you can guarantee that you will get a 100% Human score when you scan your content with an AI detection tool.
Let’s learn more about how schools and universities are dealing with the issue of AI being used in the classroom.
What AI Detection tools do Schools/Universities Use?
If you are looking to make ChatGPT undetectable, you should probably know which AI detection tools your school or University uses.
These are the tools most likely that Schools use:
Winston AI
Many teachers use Winston AI to check their students’ work as it is the most accurate AI detector on the market. This AI checker has an accuracy rate of 99.98% when checking for AI-generated content.
If you want to make your work undetectable you need to work with the best AI content detection tools and Winston AI is the best AI detection tool on the market.
I asked ChatGPT for an essay and then scanned it with Winston AI.
As you can see on the right of the image above, the text was correctly identified as AI-generated by getting a score of 0% human.
We have also found this to be the case when testing “undetectable” tools like Undetectable AI and StealthGPT.
Turnitin
Turnitin is a plagiarism detector which also now detects AI content. They have partnered with many educational institutions to test whether an assignment is original and if it was generated by AI.
Unfortunately, there is no way for students to check if their work will be marked as AI before they hand it in. That’s why you should always check your assignment with Winston AI first before you submit it.
Let’s learn how you can try to make text generated by ChatGPT undetectable.
3 Ways to Make ChatGPT Undectable
There are at least 3 ways you can try to make your ChatGPT text bypass AI detection.
Undetectable Tools
The first way is to use an AI tool to make the text “undetectable”. These tools are also called “humanizers” as they intend to make your writing more human-sounding. This name doesn’t make any sense at all if you understand how the tools work.
These “undetectable” or “bypass” tools change the word order and grammar of a text so that it is more random. They basically add capital letters where they shouldn’t be and use really strange patterns of language.
These humanizing/undetectable tools can only bypass inferior AI content detection tools so it is pointless even trying this. Not only that, but the output generated ends up being terrible quality. Overall a waste of time.
Improve your Prompting
Another way to attempt to make your AI text undetectable is to improve your prompting.
Prompt engineering is the art of creating a prompt to get the most out of an AI tool. Your prompt should make it crystal clear what it is you need.
I have tried many prompts to bypass detection and it really depends on the AI detection tool. For basic AI detection tools, you can just say something like, “Write in the style of Mark Twain” and you will probably make your text undetectable. This won’t work for more advanced AI detectors like Winston AI.
Like the humanizing tools mentioned above, this doesn’t work with top-notch AI detection tools so, again, not worth your time.
The Legitimate Way – Edit Yourself
The best way to make your ChatGPT undetectable is to edit it yourself.
We all know that AI tools like ChatGPT exist so why not just allow them under certain circumstances and allow students to use them and improve the output?
The result from an AI tool should be the baseline for an assignment and students should be expected to write something significantly better. It’s like being asked not to use a calculator and then going out into the real world where everyone has a calculator on their phone.
Use ChatGPT as a guide only
The best way to make chatGPT undetectable is to only use it as a guide.
You can use ChatGPT to brainstorm ideas for your work. ChatGPT will often give you more ideas that you didn’t think of yourself.
You can also use ChatGPT to help you outline your work.
But, you should write the text yourself. This is the only way to guarantee that your work will be undetectable.
A Word of Warning about Using AI
Many professors are warning that students use AI to do their work for them and end up learning very little.
Since the dawn of the internet, there will always be those students who spend more time trying to do something the lazy way instead of actually doing the assigned project.
AI can make us lazy and lose skills like writing, which are super important.
No matter how good AI gets, you will still need writing skills to express your thoughts in a coherent manner. In fact, people who write well will be more in demand as we work more on the ethics and values of AI in the modern world.
Writing and Speaking(like any skills) need practice and relying on AI to do everything will be harmful in the long run.
A Misnomer – Humanizing Text
Due to the accuracy and widespread use of AI Detection tools like Winston AI, a new market of tools that are called “humanizers”. These tools claim to make ChatGPT sound more human.
The truth is that the current tools just rearrange the text so in an attempt to bypass detection(they don’t work with Winston AI, I have tried all of them).
But, the next question is, Can you make the ChatGPT sound more human?
It all depends on the context really.
The fact is that most text generated by an AI tool is becoming impossible to spot just by reading it.
AI is most helpful when creating text that is quite boring but necessary. An example could be a topic like pricing on a website. People often don’t need an interesting story about why the price is the price(though that might be fun!).
On the other hand, AI is not so good at expressing emotion(obviously!) and experience in situations where a good story or hands-on experience is needed.
What we really need are good human writers and good AI tools(and know when to use each one in the right context).
Final Thoughts
Making ChatGPT undetectable is a very difficult task and not worth trying. If your boss or teacher has a good AI detector, the AI detector will still know that the text was written by ChatGPT.
I suggest you write the assignment yourself(Much harder, I know!). You will learn more and improve your writing skills.
