A Word on AI

Like it or not, AI, or at least what we call AI, is here to stay. It’s been a part of our lives for quite a while now, in more subtle ways such as spell checkers and Alexa. But truly smart AI doesn’t really exist at the moment (or at least not to the general public), which is why I’m not too impressed by tools such as Chat GPT or Bard.

I was briefly infatuated by Midjourney and other AI generative art apps, but the allure wore off fast, especially when I learned they were trained on the real labor of humans (often without consent and riddled with copyright logos).

And now people are flooding the book market with completely AI generated words and images, many that are confirmed to be inferior and/or bland, compared to actual written books. I, like many other readers (and authors) are unimpressed, as aside from the “AI” leaning on popular media to complete its task (what we call plagiarism), I just don’t see the point in having a machine spew out an entire book for you. Or make art in general, really.

Well.

Capitalism kinda necessitates that need, but you still gotta promote and find an audience willing to pay. Books don’t sell themselves, unfortunately.

Now, I’ve actually dabbled in machine learning, as someone who performed micro tasks online for compensation. I’ve worked for researchers on sites such as Mturk, Prolific Academic, Neevo, Telus, and others.

I’ve trained AI to identify and label images such as landscapes, cars, shoes, etc. I’ve even rated AI generated speech and recorded myself repeating phrases for AI assistants. Most recently, I’ve read transcripts of AI chatbots and pointed out issues with sensitive content. And, for a short period of time, I even played with the AI friend app Replika.

So I have an idea of how it all works, but I’m hesitant to fully embrace AI. It’s the newest shiny toy everyone (mostly tech bros) is flocking to, but it has very little regulation in how it’s being trained. Right now, the biggest issue is with retailers allowing it to scrape copyrighted content to “train” it, rather than hiring a dedicated team to feed it content directly.

Another issue with AI, as seen often in the hiring process, is that the bias of its handler gets baked into the code. If you’re a black person in the US looking for a job, you know what I mean. If your name isn’t traditionally white and American sounding, you get auto-sorted out of applications. Same for other undesired keywords that come up in resumes. And funnily enough, these same measures screen out white applicants.

Take my mother-in-law, for example; she has blonde hair and blue eyes, but because her Sicilian surname sounds “exotic,” she gets passed up for jobs she’s more than qualified for. And even when she gets an interview, the interviewer is shocked that she doesn’t meet their expectations. She’s asked, and the usual response is a sheepish, “Well, I expected a Brazilian or someone middle eastern.”

So AI has been a thorn in the side of millions for years, but now it’s going to have an even bigger impact as it’s being openly embraced by competing companies. By the way, this doesn’t just affect books or assistants, but healthcare (look at all the companion self care apps or the recent fiasco with the eating disorder chatbot), real estate agents relying on AI chatbots, and many others.

I’ve gotten a bit off track here from my original point, which is to say that I don’t use AI to write my books for me. I do use AI to help edit my books. Google Docs helps put out missing or incorrect words, but ProWritingAid takes that further by alerting me to overused or complicated phrases. It even suggests ways I can improve my text, and there’s now an option for it to rewrite sentences.

I rarely use the suggestions given because it either makes the text too formal or just...doesn’t make any sense in context. I also don’t use the new option as I’m not convinced it can do a good job. Plus, I doubt it can improve upon the text or replicate my weirdness. I similarly struggle to accept some suggestions from human editors, ahaha! I’m stubborn!

And while I would love to put out several books in a year, I am a slow writer. I also have ADHD and that makes finishing something difficult. Just look at my mountain of WIPs and ideas. I can maybe publish three to five books a year, mostly made easy by writing serials / publishing chapters serially online. And then there’s the issue of book covers. I’m really picky and don’t like most AI “art”.

There’s also the issue with most of it catering more to white audiences, so I’d have a hard time finding cover art with POC (an issue I currently have), and cover art gets pretty expensive! I spent $500 on a custom cover for I Think of You Often, and I would do it again (I will for the second book, especially), but I don’t always have the money for it.

All this to say that my books are written by a real person (moi). I really hate that it even needs to be said, but with the state of the world right now, I kinda have to. It would also be great if people making AI content actually gave a disclaimer, but I’m a skeptic.

I write because I’m a creative person. I’ve always been fascinated by art, particularly books, which my mom exposed me to at a very early age. I was a hyperlexic child and could read before preschool. Writing later became a passion of mine to help clear the ideas and daydreams swirling from my head, but also a means to redirect my depression and anxiety into something productive.

Something to process trauma...that’s why I put so much of myself into my writing. My lived experiences, my rocky childhood, the difficult relationships with my family...

Even if AI generated content takes over the book world, even if my sales dry up...I’d still write and publish in hopes of building a connection with people. With people who can relate to my experiences and want someone they can actually reach out to (but not physically touch; I am not a hugger).

Because I am an artist and I live to create. It’s not all about the money.