Notes · Dispatch №011
The most expensive way to sound like everyone else.
A meme about asking a frontier model for a LinkedIn post, and the uncomfortable rule it illustrates: AI multiplies the quality of the question.
I asked the most powerful AI on the planet to write my LinkedIn post.
It analysed my industry. Researched trending topics. Crafted a compelling hook. Optimised for engagement. Polished tone and structure. Consumed two million tokens. Status: mission accomplished.
Then it handed me “7 Signs You’re Underestimating Your Potential (And How to Fix It)”.
Frame: the handshake · Two million tokens, one listicle
Two million tokens, and the result reads like every post you scrolled past this morning. That is not the model failing. That is the model succeeding at a lazy ask. Give a frontier model a generic request and it returns the global average of LinkedIn, at premium rates.
AI multiplies the quality of the question. Vague in, template out, invoice regardless.
The last dispatch argued that model choice is a routing problem: match the tier to the cost of being wrong. This is the other half of the discipline. Routing decides what you pay per token. The ask decides whether any number of tokens can save you. A frontier model with a vague brief is a principal architect assigned to “make it better”. You will get something polished, confident, and interchangeable with everyone else’s.
The fix costs nothing. Say who the post is for. Say the one thing it should claim. Say what it should never sound like. The same two million tokens, pointed at a sharp brief, produce something only you could have published.
The meme above cost zero tokens. I made it myself. Draw your own conclusion about who is underestimating whose potential.
Developed in the darkroom · Dispatch №011 · Fixed 12 Jun 2026
Comments
Loading...
Leave a comment
Thanks. Your comment is in.
This dispatch is open for letters
"A PMO ritual you secretly love, and the one you would burn tomorrow."
Next dispatch
Get the next frame as it develops.