By Aditya Sahai · 2026-10-21
Open LinkedIn. Scroll for thirty seconds. Count how many posts start with "Here's the thing nobody tells you".
That's not a coincidence. It's the math.
The machine is built for average
A language model works by predicting the most likely next word, again and again. Ask it for a hook about productivity and it hands you the hook that best matches everything it has read about productivity. In other words: the middle.
That's a feature, not a bug. The middle is safe, correct and readable. It's also exactly what everyone else got when they typed the same prompt.
So when ten thousand people ask for "a viral post about discipline", the internet gets ten thousand versions of the same post. Same rhythm. Same three bullet points. Same closing question.
Output got free. Judgment didn't.
The cost of making a draft fell close to zero. The cost of knowing which draft deserves to exist did not move at all.
That's the whole shift. The skill used to be making the thing. Now the skill is choosing, cutting and adding the part the model could never know.
💡 If the AI could have written it without you, the reader can get it without you too.
Question 1: What do I know that the model doesn't?
The model has read the internet. It hasn't lived your Tuesday. It doesn't know the client call that went sideways, the number you saw in your own dashboard, or the conversation with your dad that changed how you think about money.
Before you prompt anything, write down one specific thing only you know about the topic. A moment, a number, a mistake. That becomes the spine. The AI can help you shape it, but it can't invent it.
Question 2: What would I cut?
AI drafts are generous. They explain everything, hedge everything and end with a summary nobody asked for.
Taste is mostly subtraction. Go through the draft and delete every line that would be true in anyone's post. What's left is usually a third of the length and three times as sharp.
Question 3: Who exactly is this for?
"Founders" is not an audience. "A founder whose ads worked last year and stopped working this year" is.
When you name one specific reader, the language changes on its own. You stop writing for the feed and start writing to a person. People can feel the difference, even if they can't explain it.
The move
Use AI for what it's brilliant at: research, structure, options, speed. Then do the part only you can do: add the lived detail, cut the average, aim at one person.
That's the gap between AI content and creative work made with AI. One is free. The other is worth following.