Why your AI-written content sounds like everyone else's (and how to fix it)

Most people blame the robot when their AI-written post flops. The real problem usually sits one layer deeper: they fed the AI nothing worth repeating in the first place.

Nicolas Cole, a nonfiction writer and ghostwriter who's built a career writing on the internet, has spent over a decade studying exactly why some AI-assisted content lands and most of it disappears into the void. His answer isn't "write less with AI" or "write more with AI." It's a framework for understanding what kind of content you're actually creating — because not all writing carries the same value, and AI treats each type completely differently.

Here's the breakdown, and how to use it so your own writing doesn't turn into slop.

What are the three tiers of content?

Cole separates commodity content, personality content, and original content. Commodity content helps beginners, and it stays useful as a first step. Personality content and original content raise the long-term value of your library.

Think about what each tier actually looks like in practice:

Commodity content is the stuff that's already been written a thousand times. "5 tips for better sleep." "How to write a cover letter." It's useful — especially to beginners who need the basics — but it has almost no scarcity. Anyone can produce it, including AI, which means it commoditizes fast and loses value even faster.

Personality content takes a commodity topic and runs it through a specific human lens — your stories, your contradictions, your way of seeing things. Same topic, completely different asset, because nobody else has your specific experience to draw from.

Original content is the rarest tier: ideas, frameworks, or language that hasn't existed before you said it. This is where a point of view creates the moat, because copycats must wait for your next idea.

The mistake most people make with AI is staying stuck in tier one forever. They ask ChatGPT for "10 tips about X," publish it, and wonder why nothing sticks. AI is excellent at commodity content because commodity content is, by definition, pattern-matched from everything already written. That's exactly what a language model is built to do — remix the existing pattern. It's also exactly why that output feels hollow to readers who've seen the same pattern a hundred times before.

Why does commodity content burn people out?

Commodity Content — keyword-stuffed, SEO-optimized blog posts; generalized How-To advice; technical walkthrough guides; medical questions & answers; etc. — has always been the least valued form of writing on the internet, AI or not. It was low-value before ChatGPT existed; AI just made it faster to produce and therefore even more oversaturated.

That's the trap. Commodity content is genuinely useful to write early on — it teaches you what your niche wants to read, and it gets words moving. But the more of it you produce, the more you feel a specific kind of fatigue: everything you write feels interchangeable with everything else in your feed. That feeling isn't a sign you should quit writing. It's a signal to climb the ladder toward personality content.

What is a "digital brain" and why does it matter?

This is the part of Cole's framework that changes how you should actually use AI tools day to day. Every time you hit publish, you are adding a new data point to your library of content—your "Digital Brain."

The idea: your published writing — every post, thread, and article — becomes a growing dataset of "approved language" that's uniquely yours. Feed that dataset into an AI tool, and the AI can start generating drafts that actually sound like you, instead of sounding like a generic assistant. Once you've written it and crystallized your thinking, added thousands of words of "approved language" to your Digital Brain, the robots can help you leverage those ideas, infinitely.

This reframes the whole AI-writing debate. The complaint "AI writing sounds fake" is usually true — but it's not because AI is incapable of sounding human. It's because most people never gave it any real material to draw from. They prompt a blank AI model with a topic and expect personality to appear out of nothing. It won't. AI can only remix what you've already put into words. If your digital brain is empty, the output will be empty too.

How do you actually build one?

You don't need a fancy setup to start. A digital brain is just an organized collection of your best writing — the phrases you've used, the stories you tell, the opinions you hold that other people in your space don't. Practically, that means:

1. Audit what you've already written. Pull your best-performing posts, emails, and notes into one document. Look for recurring phrases, opinions, and stories — that's your raw material.
2. Write more personality content on purpose. Every time you cover a commodity topic, force yourself to add one specific story or contradiction from your own life that nobody else could write.
3. Feed that material into your AI tool as context, rather than starting from a blank prompt. The difference in output quality is immediate.

Typeshare is one platform built specifically around this idea — it's designed to help writers organize their published content into something closer to a personal language model, with templates for threads, LinkedIn posts, and newsletters that push you toward personality-driven writing rather than generic output. The company has also announced a major platform rebuild focused explicitly on helping writers build and organize their "Digital Brain" inside the tool.

If you want to see the framework in Cole's own words, his Substack post on the right way to write with AI breaks down the commodity/personality split in more detail and is worth reading directly.

Why do people hate "AI slop" so much?

Slop isn't really about AI. It's about content that has zero personality signal — writing so generic it could have been produced by anyone, about anything, for any audience. People don't hate AI-assisted writing; they hate writing that feels interchangeable, and AI just makes that kind of writing easier to mass-produce.

Here's the part that trips people up: critics of AI writing often point to one weird word choice or one clunky sentence and use it to dismiss an entire piece as garbage. But a single awkward phrase doesn't make something slop — a total absence of a point of view does. You can fix a bad sentence in ten seconds. You can't fix an empty argument in ten seconds, because there was never an argument there to begin with.

This matters for how you edit AI drafts. Don't obsess over individual word choices. Ask a bigger question: does this piece say something only you could say? If yes, polish the sentences. If no, no amount of sentence-level polishing will save it — you need to go back and add the personality or original thinking first.

What does ownership actually mean in the age of AI?

Ownership equals association. Volume builds it, and personality details make it strong. Nobody owns individual words or the alphabet — that's shared property. What you can own is the association between a specific idea and your name, built through consistent volume and specific, hard-to-copy personal detail.

This is also why your life story functions as an unmade data set — AI can only learn it from your own writing about it. Nobody can prompt an AI model to produce your childhood, your failed businesses, your specific client stories. That's the one input AI genuinely cannot generate from scratch, which makes it the most durable material you have.

How should branded and team content use this framework?

One detail worth stealing for teams and companies: commoditized content doesn't have to feel faceless. A company blog post about a generic topic can still carry a specific human voice if a real person signs off on it — "written by Tiffany" instead of an anonymous brand byline. The future of branded content isn't zero humans behind the curtain; it's showing the several humans who actually did the thinking, even on commodity-tier posts. That single move — attaching a real name and voice to otherwise generic content — pulls it up a tier without changing the topic at all.

Which content should you write with AI, and which should you write yourself?

A simple gut check before you open any AI tool:

• Commodity topic, need it fast? Let AI draft it, but assign a real person's name and voice to the byline.
• Personality topic? Write the core story or opinion yourself first, then let AI help with structure, headlines, or formatting — never the substance.
• Original idea? Write it entirely yourself. This is the material that becomes tomorrow's "approved language" for your digital brain, so it needs to come from actual thinking, not a prompt.

Humans do the thinking and the writing. Robots do the repeating and the remixing. That division of labor is the whole game. AI can repeat your best ideas at scale forever once they exist — but it cannot generate the original idea that gets repeated. Every hour spent producing more commodity content is an hour not spent creating the one thing AI can never manufacture on its own: something new, from you, that's actually worth remixing.

If you're going to publish something today, ask yourself one question before you hit send: could a competitor's AI model have written this exact piece? If the honest answer is yes, you've just handed the internet another piece of slop. If it's no, you've added another data point to a library nobody else can copy.