Craig Leonard put it in a single line on my podcast: use AI as an advisor on your profile, never as the writer, because AI-written profiles are mediocre by design.
He did not say AI writes badly. It writes fluently and never misspells anything. He said mediocre by design, and that is a claim about the mechanism. The mechanism is the problem.
Why are AI-written LinkedIn profiles mediocre?
Because a language model produces what is most probable given everything it has read, and what it has read includes several hundred million LinkedIn profiles.
Ask it for a headline and you get the statistically central headline. Ask for an About section and you get the shape those sections most commonly take: results-driven professional, passionate about, proven track record, leveraging cross-functional synergies. Those phrases are not errors. They are the average of a hundred million people describing themselves, and the average is what a profile has to escape.
A profile exists to answer one question for a stranger: why this person and not the forty others who do the same thing. An answer generated by averaging the forty others cannot do that. Better prompting does not fix it. The tool is working correctly and the job is the wrong shape for it.
What does a strong profile do that AI cannot?
It does three things, and each one requires somebody who knows you.
It picks. A career contains hundreds of facts and a profile has room for a dozen. Deciding which twelve, for which audience, toward what you want next, is a judgment call about your future. A model does not know what you want next and will cheerfully include everything you feed it.
It uses the specific. The line that makes somebody stop is almost always a concrete detail: a number, an odd combination, a thing you did that nobody else in your field did. Thirty-three years in enterprise IT followed by writing other people’s books is a strange sentence, and it is strange because it is true and specific. An AI will not invent your specifics, and it will smooth them if you supply them, because smoothing toward the familiar is what it does.
It answers what is in it for the reader. Craig’s central point about profiles is that most of them read like resumes, which are documents about the writer. A profile that works is about the reader’s problem. That flip is a positioning decision. People get it wrong on their own more than anything else, and a model has no basis to decide it.
How should you use AI on your profile?
As an advisor. It does four jobs well.
Interrogation. Paste your draft and ask what is unclear, what a hiring manager in your field would not understand, what claims are unsupported. It is a good reader even though it is a poor writer.
Alternatives. Ask for ten headline options. All ten will be mediocre and two will contain a word or an angle you had not considered, and you take the angle and write the line yourself.
Compression. Give it your rambling four hundred word About section and ask what it can be cut to. Cutting is arithmetic, and a machine is fine at it. Then rewrite the cut version in your own voice, because the compressed output will be accurate and flat.
Vocabulary from the target side. Ask what terms people hiring for a specific role search for. That is aggregate knowledge, which a model has, and it beats guessing.
What it never does is write the finished text. The moment you paste its output, your profile becomes the average, and nobody sees the average.
The two-hour version you can do yourself
If you are not going to hire anybody, this is the sequence that gets most of the value.
Write down who the profile is for, in one sentence, and pick one audience. Not everybody, and not future employers in general. A hiring manager for a specific kind of role, or a buyer with a specific problem. Everything else follows from that choice.
Then write the single sentence you would say if that person asked what you do and you had one breath. Say it out loud first and write down what you said, since the spoken version is nearly always better than the typed one. That becomes the headline, adjusted for length.
Then cut. Take the About section and remove every sentence that would be true of somebody else in your field. What survives is short, uneven, and specific. Good. A shorter profile that says one memorable thing beats a complete one nobody finishes.
The detection problem is beside the point
People worry about whether recruiters can tell. Mostly they can. The register is unmistakable once you have read a few hundred, and the tells are consistent. Perfectly balanced sentences. Abstract nouns. Three-item lists everywhere, and an absence of anything a specific human would say.
Detection is the wrong worry. Nobody rejects you for using AI. They pass over you because the profile said nothing that distinguished you, and that would have been true if you had written the same words yourself. The cost is being forgettable, and forgettable is silent.
This is the same problem I write about with books, where the stakes are higher because your name is on the cover for years. I set that out in what AI does to writing when nobody is watching.
The part people skip
Craig also said something that costs nothing and almost nobody does. Update your profile every three to six months even when you are employed, so a layoff never finds you starting from scratch.
I would put that higher than any wording advice. A profile written in a hurry by somebody who needs a job next month reads like a document written under pressure, because it was. A profile maintained calmly while employed reads like a person who is good at something, and it is ready on the day you need it.
It also compounds with how the platform now works. Craig’s other point is that LinkedIn shifted from personality-based distribution to interest-based. Scattered posting across unrelated subjects now suppresses your reach, and that everything you do feeds it: comments, connections, and messages as well as posts. Consistency on a subject is the strategy, and consistency needs the profile underneath it to say what that subject is.
What I have seen across a hundred profiles
I have optimized 100+ profiles for executives and professionals, and the pattern is the same nearly every time.
People arrive with a resume pasted into a profile. Job titles, responsibilities, and dates. It is accurate, it is complete, and it says nothing about why anybody should care. The work is almost never writing prettier sentences. The work is deciding who the profile is for, what one thing it should make them think, and then removing most of what is on the page.
The best line in a profile is usually already in the person’s head. They say it out loud in the first ten minutes of a conversation, in an aside, and then apologize for the digression. That aside is the profile. Getting it out requires somebody asking. A machine cannot do that, because it does not know what it does not know about you.
If you want that done properly, it is part of what I do, and the same principle drives everything from a profile to a book: optimizing a LinkedIn profile and writing a book that establishes authority are the same decision at different scales. Craig’s full conversation is in his episode of Leaders and Their Stories.
The Guides That Get Your Book Written, Published, and Sold
Four short, practical guides on writing, publishing, and selling your book, plus the occasional note when there's something worth your time. No fluff, no daily inbox clutter. Drop your email and they're yours.
We use MailerLite to manage our list and send these emails. Your address is used only to send you what you signed up for. We will not sell it, share it, or use it for anything else, and you can unsubscribe anytime.
Frequently Asked Questions
The book on this: Focus on LinkedIn is 276 pages on writing a profile that reads like a person, which is the one thing a model trained on everybody else structurally cannot do.