I was just thinking about this, the proof of work angle. Maybe written blogs need to come with recordings of the process now too, otherwise they will be suspicious.
What a goofy situation to imagine. I hope we can figure out a way though. I personally have no interest in reading anything spat out by an LLM, so if anything can be used to prove that an author wrote something themselves, I'm interested
It's a tough problem, since I assume any common proof-of-work standard will inevitably generate a big enough dataset to train on.
But nonetheless a problem in desperate need of solving!