Did I write this post, or did my Sensei?
You will never know. And that, on its own, already breaks the internet.
The dead internet theory is not science fiction
The dead internet theory says that most of what you read online is produced by machines, for machines, with a few humans watching from the side. For years it sounded like a conspiracy. Today it sounds like a forecast with a date on it.
Models are getting better at everything, and one of the things they are getting better at is sounding like a specific person.
Your style is already in the data
Take a good model. Give it a prompt with your full LinkedIn history, every post and every comment. Add your conversations with your friends, your emails, your messages. Ask it to write the next post.
Do you really think it cannot mimic your writing style almost perfectly?
I am not talking about fine tuning. No training run, no special model. Only prompt engineering, with context windows that already hold years of what you have written.
And this is not only about social media. It is every kind of digital communication. The email you got this morning. The message from a colleague. The comment under your post. The reply from a founder you pitched. Any of them could have been written by an agent acting on that person's behalf, in their voice, with their opinions, and you would not notice.
Execution is going to zero, writing included
We talk a lot about execution going to zero. Writing is execution too. A draft in your own voice takes seconds now, because thirty of your past emails are enough to imitate the style.
So the question "is this human?" stops having an answer you can check from the text. What is left is a different question: does this person stand behind it? A message you approved and sent is yours, whoever typed it. A message your agent sent on its own is something else, and the internet does not yet have a way to tell the two apart.
What this breaks
It breaks the idea that a post is evidence of effort. It breaks the idea that a reply is evidence of attention. It breaks every signal we used to tell a real person from a crowd of accounts.
What survives is what cannot be generated: the person you met, the call you had, the thing someone did that you saw them do. Trust moves from the text to the relationship behind it.
So, did I write this post? I decided what it says. I decided it was worth publishing. Whether my fingers or my Sensei typed it is the least interesting part.