Alright, let’s talk about AI. Or, well, not AI itself because it’s really NOT Artificial Intelligence but really good machine learning. I know there are specific terms, apologies if I miss them; and if anyone has greater expertise and wants to jump in, I’d be grateful.

AI Updates in Brief

To begin with, let’s talk about the recent changes because those are the ones that are having the biggest effects on the publishing end.

Specifically, what we’re seeing is the new ‘AI’ (and I’ll stop that right now, with more discussion of why it’s important later) software being pushed out at ever faster rates in various aspects of publishing.

In particular, we’re seeing huge improvements in:

  • audiobook AI voices
  • autogenerated content (words and images)
  • grammar and spelling and other ‘smart’ language software
  • language translation

These changes we’re seeing are coming at ever faster rates, with significant jumps in processing and accuracy. Some of these changes are much quieter, but all have the potential to be quite disruptive. Let’s tackle them in order.

Language Translation

So, this has been happening forever. It still isn’t anywhere near perfect, but the one thing I’ve noticed is the use of machine translation to conduct the initial translation and then a human editor to fix it afterwards.

This has driven prices for (simple) translation down, from the old use case of around $0.10-0.15 per word to as low as $0.02 in some instances.

I won’t go into translation in too deep of a detail; but the biggest issue with MTL is that it misses out on localisation on both ends and humor. This ranges from common phrases (we see this in particular in our titles for System Apocalypse: Australia where ‘Town Under’ which is a clever pun and local phrase  doesn’t really translate for German and thus we have to find something that evokes the same feeling) that don’t translate well to idioms and written dialects.

Add on humor, which can entirely miss in another country (different perceptions of what is funny or gross) or puns (which require a play of word in the original language) and machine translation falls down flat.

On top of that, if you’re a literary writer, with thematic moods, alliteration and the like, a straight machine translation will often miss the context clues. However, for non-fiction or straight forward text (thrillers are a great example); it is possible to go the ‘cheaper’ route.

Of course, there’s the (minor) question of copyright here; since if the translation is only minorly adjusted, is the copyright really yours (or the translators?) or is it the company’s?

More on copyright below.

Grammar and other writing aids

Another, quieter evolution of machine learning and growing sophisticated programs is in grammar and other writing aids. Some of this software is getting quite powerful, with options to change the kind of writing to be more or less formal or for fiction.

This is where writers have looked at for saving money, using this software (or softwares) to replace copy editors and sometimes line editors. Sometimes, the software is even better than some copy editors who might know grammatical rules (or more recent stylistic changes) as well as the software.

I’ve noticed a number of proponents in various groups advocating the use of such software, on the first pass if not the only pass.

There are still problems. You can’t expect the software (right now) to look for plot inconsistencies (blue eyes turning gray); missing words are often missed, typos into other words can often be missed. Basically, it’s like Word’s spellcheck dialed up to eleven.

But it can do things like find repetition of the same word (like the use of Something at the start of a sentence, or ‘sword’ multiple times in a fight scene without variation).

Very useful in such cases, though you often have to adjust the software to work best with your own writing style to some extent.

The problem, and I don’t think we’ll see AI software crossing /fixing these issues anytime soon is when writing crosses over from ‘correct’ to ‘serving the story’.

For example, the use of … as a way to break up a story, or to add in a dialogue trait to a character. The use of alliteration or repetition, which might or might not be on purpose to build tension. Or even the addition of commas during a sentence, when sometimes it is not necessary for clarity; but is useful for pacing.

Author voice really. And I worry, a little, that over-reliance of such software will destroy the author voice.

Of course, so can an over-zealous editor.

And we haven’t even discussed how the widespread adoption of such software might shrink the market in a really competitive market as it stands. While some people might point out that editors should just shift to line editing or even better, dev editing, I’d say no.

It’s not that easy.

The only way dev editing works is through experience. You need your dev editor to know not just story beats and plot structure, but themes, pacing, character voice and arcs and a whole lot more. Much of which needs to be trained, not just by reading, but by editing.

And the only way they learn is by doing, which is why many copy and line editors do a little of dev editing; but without clients at all, they’ll never progress. Which leads to a death spiral of few good dev editors.

Which is sometimes something that is extremely important.

Here we run into the problem of the commons. It’s ‘better’ for individual authors to make as much use of these software as possible, if it can drive their cost down. But it is worse for the writing community if we don’t give editors the work they need, because eventually the number of good editors we have will shrink, to the point where the people who are doing it are those with sufficient independent wealth to take low paying jobs to build up their skillset.

And yes, I know some (a lot) of editors are writers, but that doesn’t mean writers are good editors (or vice versa!). They’re different skillsets with some overlap.

Still, I think, we’ll see this kind of software become more and more common, but authors should go into them knowing the limitations. Understand what it not just misses, but what it is likely going to do to ‘harm’ your author voice.

This is getting long. I’ll tackle the next two (which are even bigger topics) in another blog post.

Like the business blog post? Want to support me writing more of them? Want to read ahead (2 weeks) of others? Become a Patron and choose the $2-tier to be able to read the business posts only and ask questions about the business side of writing.