Amazon and Goodreads must take steps to combat the flood of AI-generated content that will mislead readers and damage author reputations.
I see no problem with AI-generated books existing, but as is the case for any books it has to be possible to review them and evaluate their quality in the proper context.
I used to be a pretty active Goodreads librarian but that site has really gone downhill over the years. It had potential but it’s been largely squandered. Been thinking of looking into Bookwyrm but I’m worried it’s actually good and will suck all my free time away again.
It becomes a problem, if a) the market gets flooded with these things. Especially if they are pushed by a bot cartel and b) known, real author names are used, in order to increase the impact.
As long as their quality can be evaluated I don’t see the problem here, people won’t buy the crappy ones. That’s the main reason why there’s a problem with them reusing an existing author’s name, it makes evaluating the quality harder.
But who evaluates all those books? AIs can pump out thousands of books a day and the fake reviews pumping them up, too.
Just look at Amazon right now. It’s flooded with cheap chinese knockoffs, that often even have relatively good reviews. These are “artisanal” fakes, in that there’s actual physical goods involved, which creates overhead and thus limits the amount of fakes that can be introduced.
And now imagine this with autogenerated books from autogenerated authors with autogenerated reviews.
It might be possible to spot fakes. But real authors will drown in the cheap crapcontent flood.
It’s a dynamic similar to telemarketers versus robocallers or spam. Bad books generated by humans have a cost in time and effort to create them, whereas ai-generated bad books have essentially no downside for whatever organization is pumping them out. Eliminating the barrier to entry of authoring a book also eliminates consequences for failure and as the cost to do a thing drops towards zero the frequency of that thing getting done will skyrocket. Eventually the signal to noise ratio becomes too poor and people simply reject the medium.
The theory of infinite monkeys with infinite typewriters and infinite time focuses on the production of the works and kinda ignores the poor bastards who have to read that shit.
It might help to imagine it as if it were a different product. Imagine if you went into a shop to buy a computer and there were thousands of different computers for sale. Some of them made by actual people and some made by AI. The shop doesn’t test these computers so you might buy one that doesn’t work, or is missing a vital component, or is just a case full of sand. Each one has reviews but they’re all rated 5 stars with AI generated review text. What do you buy?