After I finish writing an article I have the temptation to send it to Claude.
I do this because I think Claude will give some feedback that will “improve” the post I wrote.
Maybe it will catch some grammar errors. Or find something that is not accurate.
I used to feel good when I got the approval from Claude that it is a good article.
But what Claude think is good or bad is not the word of God.
If it says it’s good it doesn’t actually make it objectively good.
It’s just something that the LLMs spins out.
There is no objective function to measure the goodness of the article.
You write to express yourself.
Not to express what Claude happens to respond to your article.
Be raw. Be original.
Who cares if you make some grammar mistakes.
People are capable of understanding your writing with grammar errors.
Stop being the mouth peace of an AI.