How to Bypass AI Detection and Effectively Reduce the AI Rate of Your Articles

In this digital era, AI creation has become a trend. Whether it is writing articles, drawing or designing, AI can produce works quickly. Many people have used AI to create works, but they will always encounter the situation that the AI rate is too high. We have talked about the AI rate many times before. Today I will tell you some tips to reduce the AI rate.

1 How does AI rate come from?

AI rate is actually the probability value that a text is considered by the algorithm to be generated by AI. It is also called "Ai probability", "Ai detection score" and so on. Many schools, journals, and even recruitment units have begun to require "Ai test reports" of papers, resumes, or opening reports. Most mainstream detection software judge whether a passage is written by AI from multiple perspectives, including word distribution, grammatical structure, rhythm fluctuation, etc. in short, AI rate detection does not look at whether the passage is copied from others, but how similar the passage is to the article generated by AI. This is from the style.

2 Ways to reduce AI

In the AI creation scene, "reducing AI rate" is not to stop AI, but "human-computer cooperation", which weakens the traces of the machine to the point where it can no longer be weak, so that the content is fast and "human". The following methods are applicable from the beginning of AI generation to the final draft. No matter what kind of scene, if your AI rate also exceeds the standard, you might as well try.

2.1 Add personalized qualifier

High risk AI rate general instruction: "write a short article about young people's health preservation".

Optimization instructions (low AI rate oriented): "from the perspective of 25-year-old workers, health preservation should include details such as" drinking wolfberry tea while working overtime and being teased by colleagues ". Colloquial words such as" we "and" Laolao "should be used instead of" to sum up "and" this shows ".

AI will have a default "safe and universal" output. If "identity perspective+specific scene+language taboo" is added, it is equivalent to setting AI in a cage that cannot be used for high-frequency sentence patterns, so as to generate content closer to human expression.

2.2 Let AI output by task

First assign ai the writing task and let AI write the outline of the whole article (including the theme of the article, divided into several paragraphs, and the general content of each paragraph), then issue instructions to AI respectively: "generate the text of paragraph 1/2/3 according to the outline", and finally use the prompt words such as "add some examples or arguments to the following paragraphs" to guide AI to add corresponding cases in the text paragraphs.

• Let AI list three outlines in different styles, such as "story opening", "data opening" and "question opening", choose one and then change the logic, like interleaving "cause analysis" and "case" to confuse the fixed structure of "first cause and effect" of AI.
• For each subheading, AI should only write the core content of 200 words, and "each paragraph must have a rhetorical question or exclamation sentence";
• Finally, let AI add "three new cases of young people's health preservation (such as' morning C and evening a turn into morning tea and evening bubble ')" and then insert them into the article.

These split AI content can not generate "mechanized long text", and it is easier to add personal ideas in the later integration.

2.3 Add real life details

AI can only write "conclusion", not "details", which is the simplest detection method. So after we write an article with AI, we can add some of our own experiences, friends' stories, and even videos we usually brush into the article.

2.4 Increase contradictions and disputes

The content generated by AI is often "smooth and stable". For example, "young people should take exercise and diet into account in their health preservation", but human beings have "tangles and preferences". Appropriate expressions such as "I think", "but I have tried" and "instead" are added to show the "sense of contradiction" of thinking, which is more like the real idea of human beings.

2.5 Breaking the logic inertia of AI

The sequence of AI paragraphs is generally "background → problem → reason → solution → summary".

For example, write "young people lose their hair". AI structure:

More and more people lose hair ->harm of hair loss ->causes of hair loss (stress, diet) ->solutions ->summary.

Then you can change its structure:

1. my roommate lost his hair recently, and lost a lot of hair every time he combed his hair (at the beginning of the case)
2. in fact, she is not the only one. I checked the data. Before the age of 25, 30% of people began to lose hair (background)
3. it was useless for her to eat black sesame at first, but later she learned that it was caused by staying up late (reason+counterexample)
4. now I go to bed at 11 o'clock every day and use a massage comb

Replacing "dry background" with "case" and then disrupting the sequence of "reasons first and solutions last" are more in line with human "story telling" thinking.