Everybody’s writing with AI now. Whether they admit it or not, it shows.
You can see the little tells everywhere: the strangely inverted sentence, the oddly formatted paragraph, the over-polished neutrality, the sudden obsession with em dashes. What used to be the hallmark of a particular kind of writer has become the default setting for anyone with a prompt box and a deadline.
The mistake people make is treating this like a moral problem. It isn’t. AI-assisted writing is here. It’s going to stay.
The real question is whether you use it to publish clearer thinking or to publish generic noise faster.
And there’s a commercial awkwardness we need to acknowledge: people still devalue work when they believe a machine produced it.
You can argue that’s irrational, and you’ll be right. But it’s still true.
For now, the “I used AI” disclaimer often reduces perceived value, even when the underlying ideas are solid. That cultural glitch will fade, but it’s part of the current landscape.
So let’s be practical. If you want to write with AI and still sound like a human with a point of view, you need to understand what writing actually is.
Words were the cost. They aren’t anymore.
For most of modern history, writing was expensive because producing words took time and effort. Fingers on keys. Pen on paper. Drafts. Rewrites. Fatigue. It made sense that writers charged by the word because the word was the visible unit of labour.
That economic model was already wobbling before AI.
SEO and content marketing didn’t just change how we publish; they changed what we considered “good.” The blog post went from being a compact vehicle for clarity and opinion into a bloated factory output designed to rank. A lot of it became lifeless: long, repetitive, and written for machines first.
Then distribution followed the same depressing path:
Social platforms throttled organic reach unless you pay.
Email became a crowded, low-attention battleground.
Video grabbed the algorithms because it holds attention, not because it always transfers knowledge best.
Publishing platforms built walls, subscriptions, and closed loops.
People still want useful ideas. They’re just exhausted by the way those ideas are packaged and pushed.
That’s why I think blogs (and direct, human publishing) are coming back.
Not because we miss 2012. Because the economics are cycling.
When every platform turns into a toll road, the advantage goes to people who can build attention with genuine value and then monetise around it through services, products, consulting, premium tiers, software, community—whatever fits.
AI accelerates this return to fundamentals because it makes one thing obvious:
Words have become cheap. Original thinking hasn’t.
Two parts make great writing, and only one is “writing”
Strong publishing has two components:
The fuel: ideas, evidence, experiences, data, judgement, taste.
The vehicle: structure, clarity, pacing, and language that lets the reader absorb meaning quickly.
AI is getting brilliant at the vehicle. Drafting, restructuring, summarising, tightening, changing tone, creating variants—these are machine strengths.
But if you give AI weak fuel, you get weak output. The tool can only remix what it has seen or what you supply. That’s why so much AI writing feels the same: it’s based on the same average inputs and the same safe patterns.
So the real competitive advantage is moving away from “being able to write” and towards “being able to think, notice, interpret, and create.”
That sounds obvious. It’s also the part most people are avoiding because it is harder than typing.
The three-step method that actually works
If you want to use AI without publishing the same grey sludge as everyone else, use this workflow.
Step 1: Do the human job first (ideas, evidence, angle)
Stop starting with “write me an article about X.”
Start by collecting material that the machine cannot authentically generate:
A sharp point of view you actually hold.
A specific observation from your work.
A story with details that demonstrate the point.
A piece of data you found, analysed, or questioned.
A contradiction in your industry that deserves daylight.
A “this is what people get wrong” argument with a clear stance.
This is why the future belongs to people who look more like journalists and creators than “writers.”
The value sits upstream of the prose: gathering reality, finding signal, forming judgement, building perspective.
It’s also why reading and learning remain non-negotiable.
Everyone has a limit. You get tired. Your attention collapses. You can’t consume infinite content. If you want to produce anything worth reading, you have to do the work of turning the world into insight.
Step 2: Extract your thinking quickly (messy is fine)
Most people try to produce polished prose from the first line. That encourages safe writing and kills originality. Better results come from separating extraction from presentation.
Extract first. Shape second.
Extraction can be voice notes, a rough bullet dump, a messy argument typed at speed, a transcript from a meeting, a rant you would never publish. Messy is acceptable because its job is to capture thinking while it’s alive.
The more you can reduce friction at this stage, the better. Typing is a slow interface for many people. Speech is faster. The tools for transcription keep improving. The point isn’t perfection. The point is getting your real material out of your head and into a format you can work with.
Step 3: Use AI like a police statement-taker, not a ghostwriter
This is the analogy that matters, and it’s cleaner than most “AI writing” metaphors.
A police officer doesn’t sit someone down and say, “Please write a perfect statement in flawless English.” They interview. They ask questions, clarify details, establish a timeline, identify what matters, remove ambiguity, and then turn a messy human account into a structured statement that another person can understand.
That process has three clear roles:
The witness provides the account (their words, their perspective, their detail).
The officer provides the structure (sequence, clarity, relevance, completeness).
Accuracy is checked (read-back, corrections, confirmations).
That is how you should treat AI.
You are the witness. AI is the statement-taker.
Your job is to supply the reality: what you saw, what you believe, what you learned, what you can justify. The machine’s job is to organise it so someone else can consume it.
This is the practical shift that stops AI writing sounding like AI writing. You stop asking for “an article.” You give it an account and demand a professional conversion into publishable clarity.
Here’s what that looks like in practice:
1) Give the account. Paste your transcript, notes, data, examples. Do not hold back the messy bits.
2) Make it interview you before it drafts. Ask for clarifying questions and gaps:
“What is vague or unsupported here?”
“Where is the argument jumping steps?”
“What’s the strongest example and where should it go?”
“What would a sceptical reader push back on?”
“Which points repeat and can be merged?”
3) Draft, then run a ‘read-back.’ Make AI prove it understood you:
“Summarise my argument in 5 bullets. Confirm each is in my notes.”
“List any assumptions you made that weren’t explicitly stated.”
“Highlight lines that feel generic or could apply to any industry—tighten or delete them.”
“Rewrite this to match my voice constraints: shorter sentences, fewer fillers, less corporate polish.”
This approach gives you speed without sacrificing distinctiveness. It also keeps ownership where it belongs: with the human who has the knowledge and judgement.
The reason AI content feels flat is not because AI is incapable of producing good language. It’s because people are outsourcing the part that makes writing worth reading in the first place.
A note on style: the “AI tells” are fixable
If you’ve read enough AI-generated content, you can spot patterns: safe abstractions, symmetrical phrasing, predictable transitions, and rhetorical structures that repeat like a template.
You fix this by being explicit about constraints. Treat style as something you engineer, not something you hope for.
Build a short checklist and enforce it every time:
Keep sentences shorter than your AI defaults.
Remove filler lines that say nothing (“In today’s fast-changing world…”).
Cut generic summaries. Replace with specific claims and examples.
Limit certain constructions you’re tired of seeing.
Maintain your vocabulary. Avoid the borrowed voice of “professional internet writing.”
Style is not decoration. It’s the delivery mechanism that determines whether your ideas land.
The uncomfortable part: distribution will dominate
Once production becomes easy, distribution becomes the hard part. That’s already happening.
Organic reach is fickle. Algorithms prioritise what benefits the platform. Large followings can be strangely hollow if the feed doesn’t show your work. Meanwhile, the amount of content keeps rising.
So “writing with AI” has a second half: marketing.
You need a distribution plan that matches reality:
Consistency in format and point of view so people recognise you quickly.
Clear audience focus so the right people stick around.
A cadence you can maintain without burning out.
Repackaging for multiple contexts (newsletter, posts, threads, short summaries).
Collaboration, PR, communities—anything that earns attention without begging the algorithm for mercy.
Paid distribution where it makes sense, because sometimes money is the shortest path to certainty.
This is where the real moat starts to form. As the cost of creation collapses, the value of attention rises.
Where this ends up
If you spent years improving your writing craft, it still matters. Clarity matters. Taste matters. Voice matters.
But the moat has shifted.
The advantage now sits with people who can:
generate original ideas and useful insight,
gather evidence and interpret it,
convert messy thinking into clean communication with the help of AI,
distribute that communication effectively.
AI makes the writing cheaper. It does not make your judgement cheaper. It does not make your experience cheaper. It does not make your curiosity cheaper.
In the next few years, the winners will fall into two overlapping groups: people who can create signal, and people who can distribute it. The smartest will build competence in both.
If you want to write with AI and keep your voice, keep the ownership of the account. Let the machine do the structuring.
Then market the work like you respect it.
Andrew Holland
Originally published on LinkedIn on 14 December 2025.