Writing human-like content with AI is no easy task. I tested 74 different prompts to try to get ChatGPT to sound more human.

Most of these so-called humanizing prompts were a complete waste of time, but eight of them were absolute game-changers and have become part of my permanent prompt set.

These eight prompts can be categorized into:

1. Personas
2. Personal Experience
3. Fact Insertion
4. Perplexity
5. Burstiness
6. Unfluffing (Removing Fluff)
7. Flavor
8. Audience Targeting

When combined together, no one will have any clue that your content came from AI.

In this story, I’m going to show you exactly how to make ChatGPT sound human—not just any human, but whoever you want it to be, whether that’s yourself or your favorite copywriter. And make sure to stick around because you’ll be able to get all these prompts and start generating human-like content today!

Introduction

Google’s helpful content guidelines list requirements like original information, personal experience, and easy readability as being important—all factors that ChatGPT doesn’t do well out of the box unless you give it the right prompts.

Testing the Prompts

Now real quick, I want to explain how I tested all these prompts because grading how “human” content is can be pretty subjective. To get around this, I assembled a team of 30 viewers. Each time I tried a new prompt, I’d ask them to rate the likelihood of the content being 0 (strongly AI) or 10 (strongly human).

The top eight prompts made the cut, but to make sure the system works and my testers aren’t insane, let’s establish a baseline by seeing how they rate a piece of content that came straight out of ChatGPT.

I used my paid account and asked ChatGPT to write me a 1,000-word article titled “Beginner’s Guide to SEO.” Here’s what it came up with:

Sorry, I just threw up in my mouth a bit. "In the realm of" is a dead giveaway phrase that AI always uses, along with “navigating the ever-evolving landscape” and “bespoke strategies.”

There’s no expertise or emotion in this content—it’s full of fluff. The paragraphs are all the same length, and there’s no life to it at all.

In my testing cohort, they scored this piece a 1.3—pretty obvious that it’s AI.

The 8 Prompts to Make ChatGPT Sound Human

1. Persona Prompts: Training ChatGPT to Write Like a Specific Person

Let’s say I wanted to train ChatGPT to write like me. Here’s the training prompt:

"Here’s an article I’ve written (you can even input a Google Doc here). Analyze my tone, how I form sentences and paragraphs, the level of detail I use in explanations, the amount of humor I use, how often I ask the reader questions, the readability level I use, the vocabulary I use, and the level of emotion I convey in my writing."

I then ask it to give me a 200-word summary of what it just learned as a sanity check.

As expected, it did a great job, even picking up that I teach SEO in layman’s terms without jargon—something I consciously aim to do.

To write the content, use this exact prompt:

"Based on the writing style analyzed from the provided article, create content titled [insert your article title] at [insert your word count]. Ensure that the new content strictly adheres to the author’s writing style in terms of tone, structure, and approach as identified in the analysis."

The result? It starts with analogies like SEO being a game and includes rhetorical questions—all things I often do. There’s sentence and paragraph length variation, and no jargon. Mission accomplished! My testing team rated it a 6.9, which they’re pretty confident checks out as human writing.

2. Personal Experience: Adding Your Unique Stories and Opinions

ChatGPT might know your style, but it doesn’t know your stories and opinions. Here’s how I add that:

"ChatGPT, here’s my About page [insert your About page]. Learn about my experience and stories that qualify me as an author for the content we write."

"Also, here’s a personal opinion that will shape how you write the content: I don’t believe Google can understand content; I think it analyzes it mathematically against top-ranking articles."

When I ask it to produce content, it drops in my history (which sucked, by the way) and also adds how I feel about Google understanding content. It did go a bit overboard with the engineering stuff, so I suggest adding this prompt:

"The integration should be subtle at a level of 2 out of 10."

My team gave this content an 8.2 out of 10, which makes sense because ChatGPT wouldn’t know about my experience otherwise.

3. Perplexity: Fixing the Weird Words and Phrases AI Loves to Use

AI often uses the same weird phrases like “in today’s digital era” or “to navigate the art of SEO.” Here’s the fix:

"Don’t always use the most natural words. Use the following words fewer than three times on this page: unique, ensure, utmost. Also avoid using the following phrases entirely: meticulous, navigating, bespoke, and in the realm of."

This prompt fixes the odd language, making the content more concise and, as an unexpected side effect, more bursty (which I’ll explain next). The team gave it a 6.5.

4. Burstiness: Varying Sentence and Paragraph Lengths

Humans don’t stick to the same pattern when writing. Here’s the prompt to ensure burstiness:

"Ensure heterogeneous paragraphs. Ensure heterogeneous sentence lengths. Stick to primarily short, straightforward sentences."

The output included short sentences, long sentences, short paragraphs, and long paragraphs—all adding up to more natural writing. The team scored it a 5.5.

5. Fact Insertion: Adding Updated, Relevant Data

ChatGPT’s models are trained up until January 2022, so you’re out of luck if you need recent data. Here’s your prompt to fix this:

"Fact #1: The average salary for an SEO specialist in 2024 is $62,000 per year."

"Fact #2: As of [current month], there are 8,321 SEO jobs available on LinkedIn. Please use these facts when creating your content.”

The output now includes compelling data, and my test team gave it an 8.3. This makes sense since ChatGPT wouldn’t know these stats otherwise.

6. Unfluffing: Removing Fluff and Making Every Word Count

ChatGPT often produces fluff-filled content. Here’s the prompt to unfluff it:

"Do not include any fluff when producing content. Each sentence should provide value to the overall goal of the content piece. Strictly follow this guideline."

The result? A concise beginner overview of SEO with no fluff and actionable advice. 7.1 out of 10 from the team.

7. Flavor Prompts: Making the Content Fun and Engaging

Humans have fun when they write; robots don’t. Here’s the prompt to add flavor:

"Engagement is the highest priority. Be conversational, empathetic, and occasionally humorous. Use idioms, metaphors, anecdotes, and natural dialogue."

The output is lively, like this line: “Diving into the world of SEO is like learning to swim in the digital ocean. It might seem daunting at first, but even the best swimmers started with floaties, right?” The team rated it 7.8.

8. Audience Targeting: Writing with a Specific Audience in Mind

ChatGPT’s default writing style is generic. Here’s how to fix that:

"My target audience has the following characteristics:

1. [Insert demographic information].

2. [Insert tone preferences].

3. [Insert reading level preference]."

The content produced was tailored in tone and reading level, scoring a 6 out of 10.

Conclusion

When you combine all these prompts together, you end up with a human-sounding piece of content—maybe even better than a human! However, all this manual prompting takes time and can be subjective.

If you’re writing content on something specific like maximizing air mile points, you need to know which facts are true and what your audience wants.

If you like my effort consider giving a clap and follow for more interesting stories..

Author: Awais