Business Musings: AI: The Year in Review Part 9

I used to train radio news announcers. I trained them how to read live and on air, a skill I learned myself while still in college. In 2021, I took a voiceover class from an impressive-as-hell studio here in Las Vegas, with the full intention that I would somehow get my books into audio. I had hoped to hire a voice actor or two, or perhaps go back and do the work myself.

Life got in the way. I must say, though, that my instructors were positive about what I was doing (I hadn’t lost my chops), although I knew I had a lot to learn.

I fully intend to return to the studio, probably this summer. Getting my books into audio has been a goal since I broke up with Audible a few years back. (That’s a long story, filled with terrible editors [who were later fired], including one who disrespected me so badly that I canceled my Audible subscription in a fit of pique.) I know I have readers who prefer audio for a variety of reasons, and I’m not helping them at all by delaying my return to the format.

Years ago, in an argument with a friend, I identified myself as an audio snob, which isn’t entirely true. I do love hearing a book read well. My problem is that I trained people, so I listen with a critical ear. I correct news announcers all the time—on TV, on the radio, in a podcast, on advertisements. There are words we all learn first as readers that we mispronounce in the moment because we learned them wrong. (This is a major problem in English.) I correct them out loud when I hear that mispronunciation.

But, over the years, I realized I’m not really an audio snob, not like I used to be. I listen to a lot of podcasts—in 2x speed which isn’t how the speaker intended. I listen to some Spanish podcasts at normal speed or at half speed just so I can get the words. And yes, I’ve listened to audiobooks like that too, particularly if I think they’re “badly” read.

Why? Because I want the content. I’m not there for the performance. I’m there for the material. Which is why, when I can, I listen to the author read their own work, even if they’re terrible at it. I want to know where they emphasize a sentence, what meaning they give to certain parts of the story, what they chose to highlight.

It’s a way that I learn.

As I dithered about audio, everything changed. Joanna Penn first notified me about that almost a decade ago now, at one of our Business Master Classes. She talked about AI audio, reminded me that some sight-impaired readers used the (then) flat voice provided by online services to read the work if the work wasn’t in audio.

She challenged me. Don’t you want your readers to have that choice? I did, and I didn’t. I wanted a certain quality back then, a quality that I don’t even value myself right now.

Of course, at that point, AI audio was only a few steps above that flat reader voice. Maybe at the flat-reader voice stage. A voice robotically reading something. Readers could chose a male voice, a female voice, a particular accent, and that was about it.

Oh, baby, has AI audio advanced since then. Enough so that in 2020, I thought about using the low level of AI audio for this weekly blog, but never got around to it.

Now, though, AI audio has reached the level that I had both hoped and feared. According to an article this month in ArsTecnica, Microsoft has created a program called VALL-E…

that can closely simulate a person’s voice when given a three-second audio sample. Once it learns a specific voice, VALL-E can synthesize audio of that person saying anything—and do it in a way that attempts to preserve the speaker’s emotional tone.

There are so many ways to misuse this technology that it’s terrifying. And yet, I could make people say things they never intended back in the early 1980s. It just took a little longer. (It also took a recording on tape, a grease pencil, a razor blade and some audio adhesive tape.) What’s old is new—or at least, a lot more convenient.

Also, it’s in the hands of everyone who wants it now.

When I first came up with this year’s list for the year in review, I was going to give AI audio, and audio in general, its own special section. It really needs one. Brandon Sanderson discussed his choices for audiobooks in his yearly update. He’s leaving Audible because he calls the way they treat authors (particularly indie) “unconscionable.” He’s delivering his audiobooks through other sites. I suggest you take a look at what he’s doing, because it’s one way to handle audio.

Joanna Penn has another way entirely. She has narrated her own audiobooks for years, and she also uses AI-generated audiobooks. She marks her digital audio products in a very specific way. She makes sure that the buyer knows whether they’re getting a human voice or an AI-generated voice. She lets the buyer decide.

In May, Joanna did a long episode on her podcast that explored AI-audio in all its messy glory. She looks at the ethical issues, the ways writers can develop their own audio, and the ways they can market it.

In this year in review series, I can’t do any better than she has already done. Frankly, I’m at the beginning of my AI-audio career and she’s deep into it. So take a look at her advice.

Because I have one foot in the voiceover world (okay, maybe a toe), I’ve been following the issues with AI audio quite closely. Actors have had real problems with it for a long time now, afraid that their voices would be taken without their permission. I’ve been following SAG-AFTRA’s work on this for years. (SAG-AFTRA is the Screen Actors Guild–American Federation of Television and Radio Artists.) SAG-AFTRA is truly at the forefront of defending the rights of artists in the face of AI.

AI audio has had a seat at the audiobooks table for several years now, but the story of 2022 was AI-generated art. Platforms like Midjourney use text descriptions to produce art. The art is sometimes so good that it wins awards, when the person who ordered it up puts it in contests designed for human artists.

The problems with AI-generated art are so vast that I really can’t deal with them here. Since many of you will ask me if I use any on my book covers, I don’t. Right now, the copyright issues are not settled. Just this month, a trio of artists filed a class action lawsuit against Stability AI and Midjourney, creators of AI art generators Stable Diffusion and Midjourney, and artist portfolio platform DeviantArt, which recently created its own AI art generator, DreamUp. According to an article in The Verge

The artists — Sarah Andersen, Kelly McKernan, and Karla Ortiz — allege that these organizations have infringed the rights of “millions of artists” by training their AI tools on five billion images scraped from the web “with­out the con­sent of the orig­i­nal artists.”

The problem they’ve pointed out here isn’t just a problem with art. It exists with many AI-generated programs, from the ones who develop code to art to music. It’s a problem that needs to be settled before I can comfortably use any AI-generated art on book covers, for example.

The cases are appearing so fast that I can’t keep track of all of them. For example, an AI-generated comic book got copyright protection earlier in the year, but in December the U.S. Copyright Office opened a case to revoke the comic book’s copyright. As I said above, this will take quite a while to settle out.

In November, OpenAI released ChatGPT, which can create all kinds of text (among other things). It’s not the first AI text program, nor will it be the last. But the release created a panic among writers. It also caused a huge stir in academia, causing The Atlantic to declare that the college essay was dead and no one would learn critical thinking ever again.

Clearly Stephen Marche, the author of that essay, is an honest sort. I had a friend in college who made a living writing essays for desperate college students. They missed the exercise in critical thinking as well.

Yeah, ChatGPT would have put my friend out of business back in the day (and is probably putting other people out of business now), but cheaters cheat.

Will readers want AI-generated text? I have no idea. I’m not sure how to feel about this, though. “Creating text” is not all fiction writers do. We don’t just create text. We create stories, from our own perspectives with our own voices.

Still, there’s a worry that artificially created text will overtake regular text by creatives. Already, two months into the program, a programmer has developed a way to figure out whether a human or ChatGPT created an essay. He’s marketing a program that colleges and universities can use the way they use online plagiarism finder programs to see if a student uses their own words.

Technology battling technology. It’s truly the Wild West out there right now on all of this AI-generated stuff.

I’m waiting for it to shake out. Right now, all AI-generated programs that cross into the arts also tread on intellectual property rights as well as the livelihoods of creators. There are ethical issues for all of these programs as well as questions about the future of certain kinds of artistic expression.

What will probably happen—be it a few years from now or a decade or two—is that these AI generation programs will become new tools for creatives. Right now, a lot of creatives are using the audio programs to their advantage. I can see some of the writing programs and art programs being the same.

As I write this, I’ve also been noodling with a fantasy map generator. It’s not an AI program, but something designed to help people who are playing games. I am a terrible artist. I can’t draw worth spit. And I need a map for the Fey books.

The map of the world and its various continents exist in my head. There’s a chicken-scratched version on various sheets of paper scattered around my office. It makes no sense to anyone but me.

So I used a free fantasy map generator to communicate those chicken scratchings into something that a real artist can then use to make a beautiful piece of art.

In other words, my map is a communication tool. Most of the details on it are wrong. (I had to correct that part in an accompanying letter.) But it gets the message across.

I’m not yet sure how we professional writers could use the text tool. I see how non-professional writers can use it to get ideas across, just like I used that fantasy map generator.

I also am fully onboard with AI-generated audio products, so long as we identify them as such, as Joanna Penn is doing.

What I’m seeing, from this vantage of early 2023, is the beginning of a revolution. Where it’ll take us, I have no idea.

But the changes wrought by computers has come to a head in the past year or two. We can publish and distribute our own books. We can make a solid living at writing, if we use the tools (like crowd-funding) available to us. We can do our own audiobooks and distribute those. We don’t have to use the big e-retailers if we don’t want to.

So much has changed that it’s almost overwhelming. Scratch that. It is overwhelming.

I’m going to be spending 2023 learning the audio tool and doing some other work on things I mentioned in the previous posts. I’ll elaborate on what I’m doing on Patreon as well as in this blog—when I feel like I can safely share.  Or when I feel confident enough in what I’m doing to teach.

Right now, I’m as new to the AI platforms as most of you are. I’m monitoring the copyright battles very closely, and I’m watching what the savvy artists representatives at SAG-AFTRA are doing to defend the work of actors (and others) as these programs grow.

I’m not sure where all this is going to lead. I have ideas, but they’re partially based on ignorance. I’m not informed enough yet to trust my own opinions.

What I do know is this: 2023 will be very interesting—on all kinds of fronts. And we’ll investigate it all together.

****

As I mentioned in a previous blog, I’ve started recommending classes that we teach in this part of the post. I started this in part due to the changes in social media. But also because WMG just set up a searchable website for all of our online classes. (Before all we had was Teachable, which is not easily searchable.)

Since I don’t have anything on AI, yet, let me recommend once again Bite-Size Copyright. Understanding copyright is the best way to protect yourself, even in this new world we’re entering into.

And…as I say every week, this blog is reader supported.

If you feel like supporting the blog on an on-going basis, then please head to my Patreon page.

If you liked this post, and want to show your one-time appreciation, the place to do that is PayPal. If you go that route, please include your email address in the notes section, so I can say thank you.

Which I am going to say right now. Thank you!

Click paypal.me/kristinekathrynrusch to go to PayPal.

“Business Musings: AI The Year in Review Part 9,” copyright © 2023 by Kristine Kathryn Rusch. Image at the top of the blog copyright © Can Stock Photo / studiostoks