@dee @ariadne I never interact on AI discussions because the moral arguments won't be resolved in short text form but.. I respect both of you very much, and this is technical nonsense.

It *might* have been true before Chain of Thought and tool calls. Maybe.

But now there is absolutely no reason LLM-assisted output should be median.

Are these examples median? Are they not growing the commons?

https://bsky.app/profile/swtch.com/post/3mtwnosklvk2v
https://go-review.googlesource.com/c/go/+/822001
https://filippo.io/passkey
https://blog.cryptographyengineering.com/2026/07/29/some-notes-about-anthropics-new-results/