It is cheating to use AI to write a book?

Old problems, new tests

There is nuance in answering “Is using AI to write a book cheating?”, it’s worth doing some long walks and some thought experiments.

Start with some basics, is spell check, a thesaurus, Googling cheating? Of course not.

So we’ve established that using some ‘intelligence’ tools is ok.

So if I Google “what’s a good sports analogy for using corporate software?”, and then incorporate that inspiration into writing. Most would agree that’s fine.

But what if I highlighted some text and asked AI to find an analogy for this? Is that ok? Sounds like it’s the same as Googling, just faster.

But what if I get AI to rewrite the whole paragraph for me, or a page, or a chapter? This is what many writers pay people to do. (or in the case of photography, using Photoshop for editing a landscape photo).

Or what if I had a bot interview me and then write an entire book for me, and then publish it under my name? I tell it what I’m trying to do, the style I write it in and I share examples of things I’ve written. Isn’t that what ghostwriters do?

So are we saying it’s ok to pay a lot of money to someone to do the work that AI could do for free (and instantly), which then means only those who can afford a ghostwriter get to publish books and not others?

And finally. If the reader loved the book, their day was better for it, it was affordable, does it matter? And if AI bumped up the amount of books, affordability, quality, breadth and depth of all books, maybe it’s ok? (but opens up questions about how writers and AI training dataset sources get paid)

One thing that makes people uncomfortable is that simple questions that had clear answers in the past are now being retested because it’s so much easier, faster, and cheaper to do things.

Think of art. Once photos were not considered art and could not be copyrighted. In the 20th century that was addressed and ‘mechanical reproduction’ (photos) could be copyrighted and society started putting value on them.

Fast forward to the 21st century, you could take an ok photo, edit it in Photoshop where it resembles a far more vibrant scene, and then sell it. We were ok with it, because that process still took an hour, needed photoshop skills, and is probably something you couldn’t do yourself. But — remember that it once would’ve taken days, a dark room, playing with dyes, extremely expensive equipment and chemicals in the photo development process.

We went from modifying a photo needing days and chemicals, to an hour and software, but now jumping to a second and only needing a command. It’s making us uncomfortable, but how different is it?

Also — is it even at all realistic to debate “should we” and instead “how do we…”. Like “how do we use these tools to make better books”, “how do we produce things that AI can’t?”, “How do open up ways for great writers to publish books without traditional barriers” etc.

The questions are far more important than the answers at this point. It’s complex.