Getting in front of the AI Boom in Publishing

Years and years ago, I learned how to make ebooks.

Ebooks function like websites, comprising of code. I had enough HTML knowledge to understand most stuff, and take a guess at what the rest was doing (albeit without sometimes really knowing why it was doing what it was doing).

My former boss, who ran a small hybrid publisher, encouraged me to pursue it as a sideline, but I could see the way the technology was evolving. Software like InDesign allowed you to export files into ebook formats. They weren’t perfect incarnations, but if you knew enough you could go in and perform the fine tuning. I also knew it was just a matter of time before InDesign (or software like it) did it all.

The same applies to word processing software, such as Microsoft Word. Word already allows you to save documents as web pages, and web pages aren’t that much different to ebooks. (I will qualify that in the past, Word would bloat the hell out of the coding – I don’t know if they’ve addressed that. I doubt it, though, given they still haven’t addressed issues in Word that existed back in 2000.) There’s an online word processor, Atticus, which allows you to save and publish ebooks directly to Amazon.

Inevitably, I think the same will apply to print and audiobooks. They’ll become a SAVE AS and PUBLISH feature. I’m surprised somebody like Microsoft isn’t already working with a platform like Amazon KDP to make this happen.

When I run workshops, or talk to authors as part of my job, I always encourage them to think about tomorrow. Don’t worry about today. We know what’s happening today. Think about where technology (and trends) will be tomorrow and get in front of it.

That applies especially to AI.

As much as people don’t want to acknowledge it, AI will start performing any duty that’s currently the demesne of a computer, which also means it’s going to invade publishing (although, according to Grok, it claims AI won’t, but AI’s already shown it’ll happily mollify you with false claims).

Structural editing? Adobe Acrobat’s already doing great structural analysis. We measured its breakdown of a manuscript against a human editor’s. Our editor agreed with some things, but not others. Of course, we’d likely get the same disparities if we went to another human editor.

A lot of copyediting is what I call absolute editing. If “cathedral” is misspelled “kathedral”, then there’s only one solution to that (unless the misspelling is part of the story’s vernacular). If a sentence doesn’t have end punctuation, or dialogue doesn’t have closing quotation marks, there’s only one solution. AI will nail that with efficiency and inexorable endurance, whereas I know from my own experience as an editor that when we’re tired, we can miss things.

Stylistic preferences – such as an en dash, or em dash; or ‘single quotation marks’ or “double quotation marks”; etc. – would be easy enough to set-up. Websites and ebooks already have stylesheets, so it’s not going to be hard for AI to create and maintain one for an author.

There is some fluidity in some areas – do we separate a run-on sentence with a semi-colon or a period? Do we use a dash or a comma? But there’s leeway here. You can usually employ either option.

AI will inevitably learn and work this out just as humans would. It’ll pick up the rhythms of the author’s style, as well as voice.

The same applies to structural editing: you might want to have a conversation with a human, but AI’s simulating pretty good conversations already. People keep pointing at AI being inane. Compare it now to what it was just three months ago. It’s evolving rapidly.

Some people will be indignant. I’ll never use AI! they’ll say (and I’ve heard this often). The thing is others will.

And that’s the biggy.

In publishing, you’ll have the choice of paying an editor $8,000 for an edit that’ll take twelve weeks, or use some AI for the cost of a $200 subscription where the job will be done in twelve seconds.

While some organisations will try to remain loyal to humans, others won’t. That’s proven. Look at sectors that have been moved offshore because it’s cheaper. You don’t think businesses won’t do the same when AI saves them tens of thousands of dollars, if not hundreds of thousands of dollars? Then those holding out with their human loyalty will be forced to adapt.

Look at factories that used to be staffed by humans, and are now largely automated with only limited human supervision. I’m sure during that transition, there were people who claimed they’d stay loyal to humans, or that machines would never do the job as well as humans – how’s that look nowadays?

In fact, the simplest equation is that word processors will incorporate AI to perform the editing. Microsoft Word already has a spellchecker and shitty grammar checker. How long before those are replaced by some AI editor, Microsoft giving it some inoffensive name, like Waldo, so we view it as non-threatening? It’ll happen. (As an aside, Co-Pilot does already offer editing that’s not a part of the internal editor/spellchecker/grammar checker.)

As a species, we’ve never knocked back technology. The nuclear bomb has proven that. And, in that case, even if the U.S. had said the technology was too dangerous and they wouldn’t pursue it, do we really believe other countries would’ve shared that viewpoint? And, even if they did, do we really believe some terrorist cell wouldn’t have been working on it?

So, this is not meant to be a doom-saying post about AI. Nor is it me endorsing AI over the use of humans.

And, to qualify, I would never, ever use it for writing.

With the books I’ve published indie (Any More Complicated Than That, Prudence: A Tempestuous Night of Pervasive Desires, and the re-release of Just Another Week in Suburbia – go read them!), I’ve used and paid a human designer.

As a research tool, though? AI’s brilliant, although it can be fallible given it’s drawing from the net, and we know what a bastion of virtue and truth the net is. But (as an example) if I’m looking for an agent in the US, it can spit out a probable list for me in seconds. That used to take me hours of Googling, trawling through websites, looking at submission guidelines, etc. (A bit of advice: ask it to recheck any results it gives you, as it’ll research deeper and more thoroughly whether it’s answers are still viable.)

People keep indignantly and dismissively depreciating AI. They claim AI’s writing is inane. Or that it’s pictures are soulless and easily identifiable. And all this might be true to some extent. But we’re only seeing the beginning of the technology, and not the end of it. For our good or bad, it will improve, because that’s what technology does.

It gets better.

It evolves.

Just think about the phone you carry now compared to the phone you carried in the early 2000s to the landline you used in the 1990s. Or the car you drive now compared to what you might’ve driven decades ago. Or the Alexa you have in your house compared to … well, there never used to be such a thing.

AI will encroach more and more into components of publishing, until it plants its flag and conquers it.

It really isn’t a matter of, Will it?

It’s just a matter of, When?

So work out a way get in front of it.

Postscript: I am currently revising one of my older manuscripts. As a test, I’m going to run it through some form of AI editing to see how it goes. I’ll let you know the results.