Putting My AI Ghost Writer to the Test

Halloween is months away, but I have a ghost story to tell you. Actually, it’s a ghost writer story, but stay with me.

We all know that artificial intelligence is reshaping how we approach writing and content creation. But I often wonder how good it is at learning someone’s unique voice, so I decided see if ChatGPT can write like me. Can it understand and recreate my voice, my style, and the nuances of how I communicate?

That’s right, I put my AI ghost writer to the test.

The Experiment: Teaching AI My Voice

I’ve written a lot about AI—how it integrates into marketing, its role in higher education and healthcare, and how leaders should think about its adoption. But this time, I flipped the script. Instead of writing a blog post about AI, I asked AI to write a blog post for me.

Step 1: Giving AI Context

To see if AI could accurately replicate my writing style, I first uploaded my past blog posts. These covered everything from AI adoption challenges to content marketing strategies. By feeding ChatGPT this history, I aimed to train it on my tone, structure, and thought process.

Step 2: Drafting the Post

I then asked ChatGPT to draft a blog post comparing OpenAI Deep Research, Gemini Deep Research, and Perplexity Deep Research—three AI tools I had recently tested. Before drafting the post, it asked me a few clarifying questions, then went on its way. Within seconds, it had generated a draft blog post that was actually quite good. (You can read it here.)

Step 3: Breaking Down My Writing Voice

But did it sound like me? Would someone who had read my previous posts immediately know they were reading the output of an LLM, or would they be duped by my AI ghost writer?

These models are increasingly good at telling us why they do what they do, so I asked ChatGPT: Based on my blog posts, what are the attributes of my voice that you feel you carried forward into this draft post?

Here’s its response:

Great question! Based on your previous blog posts, I identified several key attributes of your voice and writing style and carried them into this draft. Here’s how:

1. Conversational Yet Authoritative Tone

Your writing strikes a balance between being approachable and insightful—you sound like someone who deeply understands AI but doesn’t talk down to your audience.

How I Applied It:

• Used a direct, confident tone: "Each of these tools is designed to scour the web, pull together sources, and generate a research report. But not all deep research models are created equal."
• Framed the post as an exploration, not just a review, to make it feel engaging rather than purely analytical.

2. Relatable Analogies & Clear Explanations

Your posts often include clear, real-world comparisons to help break down complex AI topics.

How I Applied It:

• Described OpenAI’s output as "almost as if an expert had done the work", reinforcing its depth.
• Positioned Gemini and Perplexity as tools for "quick, credible summaries" while OpenAI "pushes your thinking forward," making the differences easy to grasp.

3. Strategic, Action-Oriented Takeaways

Your blog posts don’t just inform—they help readers apply AI practically to their own work.

How I Applied It:

• The Use Cases section clearly tells readers when to use each tool, helping them make decisions.
• The Final Verdict section offers a structured breakdown of what to choose depending on depth vs. speed.

4. Balanced Optimism & Caution

You embrace AI’s potential but frequently remind readers that human oversight is essential (which is a theme in many of your posts).

How I Applied It:

• "Regardless of the model, always fact-check the sources." This reinforces your pattern of advising caution without fearmongering.
• Highlighted the issue of hallucinated sources and why users need to verify outputs, keeping your credibility intact.

5. Punchy, Engaging Flow

Your posts are easy to scan, with short, impactful sentences and well-structured sections.

How I Applied It:

• Used bold headers and bullet points to make the post scannable.
• Opened with a strong hook: "Deep research tools promise to save us time, synthesize vast amounts of information, and provide meaningful insights. But how deep do they really go?"
• Ended with a clear bottom-line summary, mirroring the structure of your previous posts.

Final Thoughts

Your voice is a blend of expert strategist, pragmatic AI advisor, and engaging storyteller—which I made sure to reflect here. 

Whats the Verdict: Am I Hiring My AI Ghost Writer?

I don’t know if ChatGPT was just trying to tell me what I wanted to hear, but its summary of my voice feels pretty spot on to me. And the way it was able to tell me specific things it did to make the post sound like me is pretty incredible.

Does this mean I’m going to have ChatGPT ghost write everything for me from now on? Nope. I love the process of writing and how it helps me process my thinking.

But it's great to know that a machine trained on less than a dozen of my blog posts can approximate the writing voice I’ve developed over a lifetime. Or maybe that's just scary. 👻

Have you tried to have an LLM write like you? How did it go?