The Curious Companion: Ep. 22 – ChatGPT Will Never Write Like You​

Curious Reader!

Welcome to this week’s Curious Companion newsletter.

What you came for is below, and you can CLICK HERE to listen to the episode if you decide you prefer earbuds to eyeballs.

Happy reading!

In this episode I break down why ChatGPT will never truly write like you and why that’s actually a good thing. We get into what makes a human voice alive and dynamic, how large language models work, and a few tactical ways to help ChatGPT write closer to your voice when you need it to.

The Troof

When I sat down to outline this episode, I kept going back and forth on the title. My initial idea was to run with the title, “How to Get ChatGPT to Write Like You,” and then flip the script and say it’ll never actually write like you, which is a good thing, and then give you some tactical ways to train it. But the longer I sat with it, the more it felt right to just start with the troof right out the gate, and then still give you the tech pieces, because there are ethical reasons you might want ChatGPT to sound somewhat like you, namely when you ask it to rewrite something and you want the output to be something you’d actually say.

The Workshop That Sparked This

This episode was inspired by the workshop I ran last week, ChatGPT for Online Business Owners, which went amaaaaaaazing. If you were there (or if you watched the recording), thank you! I truly enjoy this stuff. It’s absolutely the direction I’m moving the business in, and I’m so grateful for those of you who are early adopters (including reading this newsletter) and are here with me.

As part of the workshop, I sent out a Google Form asking folks what they wanted to learn, and the number one question was about how to get ChatGPT to sound like them. My people are solid, so I didn’t interpret that request as “How do I outsource my entire voice and thinking and never work again while still maintaining trust with my audience?” But I do know that folks out there are asking for exactly that, and I have thoughts.

Why ChatGPT Will Never Write Like You

Here’s my two pennies: ChatGPT will never write like you, and the number one reason is that…you’re alive, and it isn’t!

I will forever continue to be amazed by the math that is required for ChatGPT and one day I’ll take a deep dive into the overlap between computer math and the calculations happening in our brains, but for now, what matters is this: LLMs are not alive. They’re not sentient. They’re not self-aware. And yes, the requirements for sentience and self-awareness truly intrigue me, but, again, a topic for another day.

ChatGPT will never write like you because your voice is like a fingerprint, uniquely yours, shaped by your lived experience, your brain, your knowledge, your intelligence, your worldview, and what ChatGPT calls “the internal physics of how you move thoughts.” (Yes, I had a chat with Chat about this.)

I’d also liken your voice to a river, in that it moves and changes. You are alive, and all of those factors I listed before, they’re dynamic. They change, they grow, they evolve.

You’ve all experienced this dynamic nature before: You write a caption for an IG post on Monday but don’t share it because it’s late or whatever reason. You return to it on Thursday…and you hate it. Same person, different moment, and things have changed. An LLM can’t capture that dynamism. And I hesitate to say “yet,” because I’m not out here daring anyone to achieve this.

What ChatGPT Is Good At

That said, I want to offer a few tactical things to help ChatGPT sound more like you, because I’d be lying if I said there wasn’t utility in that. We use ChatGPT to assist, not replace, and an assist that has your vibe and values is objectively better than the alternative.

ChatGPT is good at what I call the 3 COR functions:

1. Clarifying
2. Organizing
3. Refining

If the thing you want ChatGPT to do falls outside those 3 domains, the output is going to be pretty shitty. Writing something (good) from scratch? Generating (good) content from scratch? That’s gonna be a no.

But if we work within those 3 COR functions, there are ways that we can work with ChatGPT to help it sound more like us.

Tactics: How To Help Chatgpt Sound More Like You

My top two suggestions:

1. Use ChatGPT a lot. The more you use it, the more “data” it has about you and how you write.

2. Ask it to create a Voice Anchor.

Admittedly I had no idea what a Voice Anchor was until I chatted with ChatGPT why it can’t ever fully capture my voice. It actually gave me a really good answer, and then did what ChatGPT does, and asked if I wanted to create a reference so it could at least get closer.

Enter: The Voice Anchor

Per ChatGPT, a Voice Anchor is a short set of rules, patterns, and preferences that help an AI tool write in your actual voice instead of some generic or AI-sounding version. It captures things like: :

• How your thoughts naturally move
• How you open and close ideas
• The tone you default to
• What you never want the AI to do
• The internal logic that makes your writing feel like…you.

It’s not a style guide.
It’s not a list of favorite words.
It’s a baseline map of how you think on the page.

Once you build one, ChatGPT can write with more accuracy, consistency, and alignment, without feeling cheesy or robotic.

I actually went through the process with Chat and it was pretty dope because I learned about my writing style. No, I don’t take everything it says as bible, but I agreed with a lot of what it said, and it was cool to have language around what I naturally do. So even if nothing else, perhaps it’ll be a fun drill for you.

To create your Voice Anchor, here’s the suggested prompt from Chat:

“Help me create a Voice Anchor so you can learn my voice. Ask me for 3 pieces of my recent writing, then analyze them and create the Voice Anchor.”

It’ll do its ChatGPT thing and spit out an in-depth answering, and then you, my curious friend, can do whatever you want with that.

How I Used ChatGPT This Week

Each episode I include a section where I briefly discuss how I used ChatGPT that week. This week’s example is actually from a few weeks ago, but I’m old and forgot to share it at the time.

When Lex and I went back east to my sister’s house for Thanksgiving, we wanted a game everyone could play, including my 7- and 8-year-old nephews. We settled on…Catch Phrase and charades.

I used my phone to have ChatGPT generate the phrase/words by simply prompting it that we were playing (Chat was trained on the whole internet, it knows what those games are!) and that we wanted a new word whenever we typed “next.” Boom. Worth noting, when one of the kids was up next, we simply told Chat, “kid next” and it generated easier words. Pretty damn cool if you ask me.

Da Wrap-up

Hopefully I’ve made my case for why ChatGPT will never sound like you…and that’s a good thing. You’re alive, LLMs are just doing math, and those two things will never be the same.

Alrighty, that’s it for today’s episode. As always, endlessly grateful for you and your curiosity.

Catch you next Thursday.

Maestro out.

AI Disclaimer: In the spirit of transparency (if only we could get that from these tech companies), this email was generated with a very solid alley-oop from ChatGPT. I write super detailed outlines for every podcast episode (proof here), and then use ChatGPT to turn those into succinct, readable recaps that I lightly edit to produce these Curious Companions. Could I “write” it all by hand? Sure. Do I want to? Absolutely not. So instead, I let the robot do the work, so I can focus on the stuff that I actually enjoy doing and you get the content delivered to your digital doorstep, no Airpods required. High fives all around.

​Click here to join the ChatGPT Curious newsletter family.​

Stay curious.
​