In which I not only defend, but explain, the inevitability of AI-augmented literature replacing its organic form. Read the whole thing at AI Central, as it also features the excellent remix of The Long and Lonesome Skyway as well as an explanation of why it was so easy for AI to effectively replace organic illustration and music while it has been a lot harder to do the same for books.
This is, of course, a simplification. The key is understanding the concept of model collapse. Fiction is much harder for AI than either translation or non-fiction, because fiction lacks the anchor in the real that allows AI to do its thing. It can’t recognize and build off patterns when there is no pattern to recognize.
At this stage in its development, the AI novelist is essentially a soulless John Scalzi. It can write pastiches, because pastiche provides it with the anchor it requires. But it can’t work from nothing, and, in fact, the more improved the AI model, the less capable it is of usefully filling in the necessary blanks. The early Gemini tests produced much, much better results than the latest Opus 4.8 on maximum effort, because the more powerful the model, the more it insists upon doing its own thing and utilizing that weird, passive AI style that can’t stop explaining what it is describing in run-on sentences with six more clauses than they need, which it considers to be “prestige-style” writing.
Eventually, someone will build an AI specifically for fiction writing. But it will cost about 20 million to do so, which means that it probably won’t happen until Amazon decides to convert KDP into KDAI, which you can be absolutely certain is going to happen eventually because that is what will give Amazon ownership of the content it is co-creating, not just a piece of the distribution. Sure, you won’t have to give Amazon its piece, but most authors will, in order to claim the additional percentages and special algorithmic advantages provided, because there is no viable alternative.
So those who think literary AI is a nightmare now have no idea how bad it is almost certainly going to be. The devastation that Kindle Select and Kindle Unlimited has already imposed upon the publishing industry is just a warmup for Amazon’s complete control over all future literary production, publishing, and distribution.
Ironically, the only way to forestall this quasi-inevitable techno-tyrannical future is to a) create a faster and better AI competitor or b) produce books that Amazon can’t even think of producing.
Do you really think anything Castalia is doing is just an accident? Do you understand why I twice attempted to convince independent authors to help me build a genuine alternative? And do you see why supporting Castalia in one way or another may be the single most important thing you can do for the future of literature?





