AI has made competent LinkedIn writing cheap.
You can generate:
- a strong hook;
- clean structure;
- tidy transitions;
- reasonable advice;
- polished conclusions;
in seconds.
That is useful.
It also changes where competitive advantage sits.
If everyone can produce competent writing, competent writing itself becomes less differentiating.
The scarce asset becomes:
- lived evidence;
- specific judgement;
- non-obvious interpretation;
- recognisable terminology;
- genuine disagreement;
- proprietary thinking.
The problem is not that AI makes content artificial.
The problem is that it can make ordinary thinking look finished.
That distinction matters.
A human-written post can be generic.
An AI-assisted post can be excellent.
So instead of asking:
“Was this written by AI?”
ask:
“Could another credible person in this industry publish this unchanged?”
That is the question The AI-Sameness Audit is designed to answer.
AI didn’t make generic LinkedIn content possible. It made it easier to finish.
Generic content existed long before generative AI.
It came from:
- corporate approval processes;
- safe brand guidelines;
- PR language;
- trend imitation;
- content templates;
- over-editing;
- executives providing no original input;
- writers without access to real expertise.
AI did not create those problems.
It accelerated them.
Give a generative model a generic prompt and it can quickly produce a polished version of a generic idea.
For example:
“AI won’t replace marketers. Marketers who use AI will replace those who don’t.”
Or:
“The future belongs to companies that combine human creativity with AI efficiency.”
Or:
“AI is a tool, not a strategy.”
None of these statements is necessarily wrong.
But what do you learn about the person saying them?
Very little.
You do not know:
- what they experienced;
- what changed their mind;
- what they actually believe;
- what they would do differently;
- what evidence produced the conclusion.
The problem is not correctness.
It is replaceability.
The AI-Sameness Audit does not detect AI
This is not an AI detector.
It cannot reliably tell you whether a human, AI system or combination of both produced a piece of writing.
That is not the objective.
The audit asks something more strategically useful:
Does this content contain enough distinctive thinking that it clearly belongs to this person or brand?
A human-written post can fail.
An AI-assisted post can pass.
The AI-Sameness Audit does not detect AI. It detects the absence of distinctive thinking.
The AI-Sameness Audit
Score the post across six tests:
- Identity
- Evidence
- Belief
- Disagreement
- Language
- Implication
Each receives:
- 0 = absent
- 1 = weak
- 2 = clear
Maximum score:
12
Call the result the:
Distinctiveness Score
It is an editorial diagnostic.
It is not an authenticity score.
It does not predict engagement.
It does not measure commercial performance.
And it definitely does not tell you how likely AI was to have written the post.
1. The Identity Test
Ask:
Could another CEO, founder or company in this category publish this unchanged?
Score 0
Almost anyone could publish it.
Example:
“Great leaders know that success is about people.”
Score 1
The post contains some industry relevance or personal context, but the author still feels replaceable.
Score 2
The argument clearly reflects this person’s:
- experience;
- role;
- market position;
- company context;
- way of thinking.
This is the fastest test in the framework.
If changing the profile photo does not change the post, the voice may not be doing enough work.
2. The Evidence Test
Ask:
What happened in the real world that caused this person to believe this?
Look for:
- projects;
- decisions;
- mistakes;
- customer behaviour;
- client conversations;
- experiments;
- original research;
- internal debates;
- observed market changes;
- first-hand experience.
Score 0
Pure assertion.
“Companies need to embrace AI strategically.”
Score 1
Evidence exists, but remains vague.
“We’ve seen this many times with clients.”
Score 2
The post contains something specific and observable.
“Across three recent implementation projects, the same failure appeared before launch…”
Now the reader has something more valuable than opinion.
They have a reason for the opinion.
Do not invent evidence just to improve the score.
3. The Belief Test
Ask:
Is there a conclusion here that genuinely belongs to the author?
Information alone is not point of view.
Compare:
Information
“AI adoption is increasing across marketing.”
Judgement
“Marketing teams should stop measuring AI adoption by licences purchased and start measuring which decisions changed because of it.”
The second statement contains a choice.
The author is telling you what matters more.
Score 0
No clear belief.
Score 1
A belief exists, but it is conventional.
Score 2
The author’s conclusion is clear, specific and meaningful.
Thought leadership begins when information turns into judgement.
4. The Disagreement Test
Ask:
Could another informed professional reasonably disagree with this?
That does not mean manufacturing controversy.
It means making an actual choice.
Weak:
“Companies should balance technology with humanity.”
There is almost nothing meaningful to disagree with.
Stronger:
“The first draft of strategy should remain human. AI should challenge the strategy after the reasoning exists.”
Now there is an argument.
Another smart person could disagree.
That is useful.
Score 0
Pure consensus.
Score 1
There is some interpretive angle.
Score 2
The post takes a clear, defensible position.
If nobody could disagree, the content may be informative without being thought leadership.
5. The Language Test
Ask:
Does the content contain language that helps this person or company own the idea?
Look for:
- memorable distinctions;
- proprietary terminology;
- named frameworks;
- original metaphors;
- repeated language;
- specific vocabulary;
- recognisable phrasing.
For example, instead of saying:
“Different stakeholders influence the decision.”
you might define a:
Decision Audience
Instead of saying:
“Balance founder and company content.”
you might create a:
Brand Carrier Matrix
Instead of:
“Choose organic posts for paid amplification.”
you might build a:
Paid–Organic Amplification Framework
The value is not inventing jargon for its own sake.
Useful terminology makes an idea:
- memorable;
- discussable;
- repeatable;
- attributable.
Score 0
Entirely generic category language.
Score 1
Some distinctive phrasing.
Score 2
The post uses recognisable terminology or framing that strengthens association with the author.
The goal is linguistic memory.
6. The Implication Test
Ask:
What should the reader think, decide or do differently after reading this?
Weak content often ends with:
“Interesting point.”
Strong thought leadership changes:
- interpretation;
- priorities;
- behaviour;
- decision criteria.
Compare:
Weak
“Brands should embrace AI carefully.”
Stronger
“Before buying another AI marketing tool, identify one decision or workflow that should measurably change. If you cannot name it, the tool is not the strategy.”
The reader now has a decision rule.
Score 0
No meaningful implication.
Score 1
Generic recommendation.
Score 2
Specific consequence for how the reader should think or act.
Calculate your Distinctiveness Score
Use this worksheet:
Test | Core Question | Score |
|---|---|---|
Identity | Could another credible person publish this unchanged? | 0–2 |
Evidence | Is there lived, observed or researched evidence? | 0–2 |
Belief | Is there a specific judgement? | 0–2 |
Disagreement | Could an informed person reasonably disagree? | 0–2 |
Language | Is there recognisable terminology or framing? | 0–2 |
Implication | Does it change what the reader should think or do? | 0–2 |
Total | /12 |
0–3 — Interchangeable
The post could probably belong to almost anyone in the category.
Common problems:
- generic claims;
- conventional advice;
- no evidence;
- no recognisable thinking.
4–7 — Competent but Generic
There is useful material, but large parts of the post could transfer to another author with little change.
8–10 — Distinctive
The reader can identify:
- experience;
- judgement;
- clear beliefs;
- differentiated language.
11–12 — Strong POV
The post contains highly recognisable thinking and would be difficult for another credible person or brand to publish unchanged.
Higher is not always better.
A factual company announcement does not need a 12.
The framework is most useful for:
- thought leadership;
- executive content;
- educational content;
- category POV.
Six signs your LinkedIn content is becoming interchangeable
The score helps diagnose the post.
These editorial symptoms help diagnose the system.
1. Universal truths
“Consistency is key.”
“People buy from people.”
“AI is changing everything.”
These statements may be true.
But correctness without information gain creates weak differentiation.
2. Predictable contrast
Patterns such as:
“It’s not about X. It’s about Y.”
or:
“The future isn’t X. It’s Y.”
are not inherently bad.
The problem appears when the rhetorical structure substitutes for reasoning.
A polished contrast is not automatically an insight.
3. Generic three-point wisdom
“Three things every leader needs: clarity, consistency and courage.”
Why those three?
Based on what?
What was excluded?
What experience produced the list?
Without those answers, structure is doing more work than thinking.
4. Frictionless narrative
Many generic posts follow:
Problem → lesson → transformation → success → neat conclusion
Real experience is often less tidy.
There may be:
- unresolved tensions;
- partial answers;
- trade-offs;
- mistakes without perfect resolution.
When those are genuine, they can make the thinking more credible.
5. Empty authority language
Watch for words such as:
- transformational;
- game-changing;
- innovative;
- authentic;
- disruptive;
- future-ready;
- purpose-driven.
The words are not forbidden.
But if removing them removes nothing from the argument, they are probably filling space rather than adding meaning.
6. No cost to the opinion
The safest content supports things everyone already supports.
Innovation is good.
Customers matter.
Culture matters.
People matter.
AI needs humans.
Leadership requires empathy.
Strong point of view usually introduces a trade-off.
For example:
“We could automate this process. We deliberately don’t because speed matters less here than preserving judgement.”
Now there is a cost to the position.
And therefore a reason to remember it.
AI is not the only source of sameness
Do not diagnose generic thinking as “AI content” automatically.
Sameness can also come from:
- over-polishing;
- safe corporate messaging;
- copying creators;
- rigid SEO templates;
- PR approval;
- trend chasing;
- weak subject expertise;
- leaders providing no source material.
AI simply makes producing a polished version of those inputs much easier.
So diagnose:
generic thinking
not:
AI usage
That leads to a much better solution.
Good AI use vs outsourcing your point of view
AI can be extremely useful in a LinkedIn content workflow.
Good AI use
Use it to:
- organise interview notes;
- summarise research;
- identify repetition;
- restructure drafts;
- generate counterarguments;
- challenge weak reasoning;
- shorten copy;
- classify themes;
- surface missing questions.
High-risk use
Be more careful when asking it to:
- invent executive opinions;
- manufacture personal stories;
- create vulnerabilities;
- generate lessons without source material;
- write thought leadership from a generic topic prompt.
The principle is:
Use AI to process thinking. Do not outsource the thinking you want the market to associate with you.
That is also why teams using AI at scale need shared editorial standards. Structured LinkedIn training and workshops can help teams distinguish faster production from stronger thought leadership.
The Source-Material Rule
For high-value thought leadership, start with something real.
At least one meaningful source should exist before the draft:
- interview transcript;
- voice note;
- meeting notes;
- customer conversation;
- internal debate;
- original research;
- case study;
- first-party data;
- executive observation;
- documented decision.
Then use AI to help process it.
Do not start with:
“Write a LinkedIn post about leadership.”
Start with:
“Here is what the executive actually said, what happened and why they believe it matters.”
Strong LinkedIn content creation should use AI to sharpen genuine expertise and perspective—not replace the source material that makes the content worth publishing.
Before and after: generic AI content
Before
AI is transforming marketing at an unprecedented pace. The companies that succeed will be those that combine human creativity with AI efficiency. The future belongs to marketers who embrace technology while staying authentic.
Audit:
- Identity: 0
- Evidence: 0
- Belief: 1
- Disagreement: 0
- Language: 0
- Implication: 0
Distinctiveness Score: 1/12
The writing is fine.
The thinking is almost entirely transferable.
After
We tested AI in a strategy workflow last month and found the worst place to use it was the beginning.
When the model created the first strategic direction, the team became faster—but also anchored around its assumptions.
We’re changing the process:
Human writes the first hypothesis.
AI challenges it.
Human makes the decision.
AI is excellent at expanding a strategic search space. I’m far less convinced it should define that space for you.
Now the post contains:
- lived evidence;
- judgement;
- disagreement;
- implication;
- recognisable methodology.
The recommendation does not need to be universally correct.
It needs to be owned.
Before and after: generic leadership content
Before
Great leaders empower their teams and create cultures where people feel safe to innovate.
After
I used to think delegation meant giving someone the task and staying available if they needed me.
It didn’t work.
They still waited for my approval because I still owned the decision.
The change was simple but uncomfortable: I stopped reviewing the decision before it happened and reviewed the reasoning afterwards.
That’s when delegation actually started.
The second version does not become stronger because it sounds more emotional.
It becomes stronger because it explains a mechanism grounded in experience.
Before and after: corporate AI positioning
Before
At Company X, we’re excited to embrace AI to deliver innovative solutions and create more value for our customers.
After
There are three things we currently refuse to automate.
Not because AI can’t do them.
Because we haven’t yet proved that automating them improves the customer outcome.
Adoption is easy to measure in licences and workflows. Value is harder.
Now there is a principle.
And principles are easier to associate with a brand than generic enthusiasm.
Proprietary language is not corporate jargon
There is a difference between creating useful terminology and inventing unnecessary names for everything.
Proprietary language works when it names a distinction people genuinely need.
A good name helps an idea become:
- easier to remember;
- easier to discuss;
- easier to repeat;
- easier to associate with its source.
Do not name trivial concepts.
Name useful ones.
Over time, those frameworks and distinctions can become part of your brand’s intellectual architecture.
Recognisable thinking matters more than recognisable writing
You can imitate someone’s:
- sentence length;
- punctuation;
- hook structure;
- formatting;
- tone.
That creates recognisable style.
But style is relatively easy to reproduce.
More defensible authority comes from recognisable thinking:
- recurring beliefs;
- decision principles;
- named distinctions;
- frameworks;
- areas of disagreement;
- consistent interpretation.
The goal should not merely be:
“This sounds like Vasileios.”
It should become:
“This is the kind of argument I expect Vasileios to make.”
That is much harder to copy.
Audit your last 10 LinkedIn posts
Take the last ten meaningful posts from your brand, founder or executive.
Score each one across:
- Identity;
- Evidence;
- Belief;
- Disagreement;
- Language;
- Implication.
Then calculate the average Distinctiveness Score.
Do not treat the average as an industry benchmark.
Compare yourself against yourself.
Then inspect:
- Which test consistently scores lowest?
- Which posts contain the strongest evidence?
- Which topics produce the clearest beliefs?
- Which phrases or distinctions repeat?
- Which posts could belong to almost anybody?
The goal is not to reach 12/12 every time.
It is to identify where sameness is entering the system.
If content production is becoming faster while the output increasingly sounds like everyone else’s, the problem may not be AI itself.
It may be that the process no longer begins with enough:
- evidence;
- expertise;
- judgement;
- tension;
- original language.
If that is the issue, a 1:1 LinkedIn strategy consultation can help define the content territories, frameworks and editorial standards your brand should actually own.
The principle is straightforward:
When competent writing becomes cheap, recognisable thinking becomes more valuable.
And before publishing the next post, ask:
Could another credible person in your industry publish this unchanged?
If the answer is yes, the next draft does not necessarily need less AI.
It needs more of you.