I write a lot here about labels for human work. Books, music, all of it. So I have to be fair and show the other side too 🙈
This year it turned into a hunt. Somebody wins a prize, few days later somebody says the text "sounds like a chatbot". And now the writer has to prove he is a human.
The Caine Prize winner this month said straight out he did not use AI, and now he sends his draft history so they can check. In September the Goncourt jury took a book off the longlist after a claim it failed an AI detection test. LinkedIn even added a "Looks like AI slop" reaction 😅
And the detectors are not magic. Pangram says 1 false positive in 10,000. A University of Maryland study from last year found it mislabeled 2% of documents. Two percent of honest people is a lot of people!!
CNN did a long one on all of this, worth your coffee time: abc17news.com
So what I think. A label should come from the writer, not from somebody's guess. You wrote it, say so. AI helped, say that too.
And keep your drafts. The messy ones, the notes, the 2am versions. Looks like that's the real proof now 😁
Get an email when Fillipe publishes