After an insightful conversation with a friend a couple weeks ago, where he explained – in a very accessible way – how artificial intelligence learns and functions, I wanted to test it out for myself.
I’m not going to dive into all the details of AI here because it’s a broad topic and I am woefully underknowledged in it, so we’re skipping right to the important stuff.
If you’d like to join a live…presentation? conversation? workshop?…I’m not sure what we’re calling it, but my friend is talking about ChatGPT with The HEA Club on March 11. If you’re not already a member, you can join live or catch the recording afterwards by purchasing a ticket.
Two of the questions I’ve heard most often from authors are:
- Is AI going to make authors obsolete?
- How does copyright work with it?
And unfortunately there is currently no concrete answer to either of those questions.
Copyright is all over the news and industry discussions right now, especially in regards to AI-generated artwork. Laws and limitations may be enacted in the future, but for the time being, it’s the Wild West out there and the best way to know what you can and can’t do with AI-generated copy is to read the terms of service of the platform you’re creating that copy on. ChatGPT, for example, says that they don’t own the copyright on anything you create with their system, but they do have the right to use it in machine learning.
Author obsolescence
That’s the real fear, right? AI is going to be so good that scam artists will be putting out a new book every three days and overwhelming retailer algorithms so that legitimate authors’ books are no longer discoverable.
I don’t think that’s going to happen. At least, not anytime soon. AI doesn’t have voice, style, emotion, or nuance yet, and it doesn’t know to break rules to make a specific impact.
Let me show you an example.
I’ve learned that the more specific or detailed your prompt is, the better ChatGPT’s output will be. So I started a conversation.
I copied over some of the voiciest posts I’ve put on Facebook and Instagram in the last week or so, including intentionally not-great grammar and some not-real words.
Me: I just posted the dumbest (but also honestest) TikTok, y’all! I want to start using “mought”… Today’s research has included UPS Stores, cisa.gov and FedRamp, and the driving distance from Idaho to California. And none of these things are vitally important to the story, I just get caught up in details. My #fridaymorningview is this chocolate cake that gave me a delicious bellyache after a walk with Arrow. My first coherent thought upon waking up was, “I want chocolate cake,” which was kinda weird because chocolate cake isn’t my go-to. The weekend is kicking off pretty nicely so far – walkies with Arrow while listening to a fabulous manuscript, chocolate cake for breaky, a group entrepreneur Q&A call right now, and plans to finish up my taxes later on. I am a wildly fun person, y’all.
The first thing ChatGPT did was basically run me through its version of Google Translate and strip away the elements that fell too far outside its spectrum of style.
ChatGPT: Sure, I can certainly try to match your writing style based on the text you’ve provided. Here’s my attempt:
Hey there! I just posted the silliest (but let’s be honest, most genuine) TikTok, guys! I’m seriously considering using the word “mought” from now on…
So, today, I’ve been doing some random research on things like UPS Stores, cisa.gov, FedRamp, and the distance between Idaho and California. None of these things are really crucial to the story, but I tend to get lost in the details.
My #fridaymorningview is this chocolate cake that gave me a deliciously painful stomach ache after my walk with Arrow. The weirdest thing is that my first thought when I woke up was, “I really want some chocolate cake,” even though that’s not typically my thing.
Anyway, the weekend is starting off pretty great so far – I took Arrow on a walk while listening to an amazing manuscript, had chocolate cake for breakfast, currently on a group entrepreneur Q&A call, and later today, I’ll be finishing up my taxes. I know, I’m basically the life of the party!
It’s…fine. Right?
An HEA member made a fantastic comparison last week when she said that AI writing is like stock photography. It works. It’s not wrong. It’s also not good.
But I wanted to give it another shot.
Me: using my writing style, please give me copy i can post on social media about the best book i’ve read lately, my plan this weekend to visit a friend and go out to dinner, my research on the best dog breeds for runners, and a romantic comedy movie i’m excited about.
ChatGPT: Sure, I’d be happy to help you out. Here’s my attempt at matching your writing style:
Y’all, I just finished the most amazing book! I’m talking laugh-out-loud funny, tear-jerkingly emotional, and heartwarming all at the same time. I can’t recommend it enough – if you’re looking for a great read, this is it!
This weekend, I’m headed to visit one of my besties and we’re planning to hit up a new restaurant in town. I’m excited to catch up and have some delicious food, drinks, and laughs.
In other news, I’ve been doing some research on the best dog breeds for runners. I’m a huge fan of running and I’m considering getting a furry running partner. I’m loving what I’m finding so far, and I think I might have found the perfect match for me!
Last but not least, I am so excited for this upcoming romantic comedy movie. I saw the trailer and it looks hilarious and heartwarming. I’ll definitely be there on opening night – who’s with me?
Again…serviceable. But it definitely doesn’t sound like Editor Jess. Our voice and style are what make us stand out in a crowded market, whether that’s a market of other authors or AI-generated copy.
So what next?
It’s here, and I believe we might as well learn how to use it instead of letting the technology get so far ahead of us that we feel left in the past. Think about your grandparents trying to program a TiVo or use a laptop. We don’t want to be them.
Could ChatGPT be useful for authors?
Yup…
Let’s generate some ideas to get unstuck in our current WIP. Remember, I’m trying to be as specific as possible to give it the best chance of returning “good” results.
Me: the heroine in my contemporary romance novel has just moved to a new apartment building and lives next door to her boss from work, who she hates. what are some funny situations i can put them in?
ChatGPT: Here are some ideas for funny situations you could put your heroine and her boss in:
Accidental run-ins: Your heroine and her boss keep running into each other in the hallway or the elevator, and every encounter becomes increasingly awkward and uncomfortable.
Noise complaints: Your heroine is loud, messy, or has an annoying pet that keeps her boss up at night, leading to a series of increasingly hilarious confrontations.
Shared amenities: Your heroine and her boss share a laundry room, gym, or other amenities, and their attempts to use them at the same time keep leading to misunderstandings and arguments.
Interference with dating: Your heroine starts dating someone new, but her boss keeps showing up at the worst possible times to ruin the mood or make things awkward.
Working from home: Your heroine and her boss both start working from home, and their attempts to stay focused and avoid each other lead to a series of ridiculous mishaps and misunderstandings.
Package mix-ups: Your heroine and her boss keep getting each other’s mail or packages, leading to a series of hilarious mix-ups and misunderstandings.
Shared interests: Your heroine and her boss unexpectedly discover that they share a common interest, like a favorite TV show or a hobby, leading to a tentative truce and some unexpected bonding moments.Remember, the key to creating funny situations is to think about the personalities and quirks of your characters, and find situations that highlight their differences and make them clash in entertaining ways.
Clearly, these are not funny. ChatGPT knows the definition of funny, it knows that funny stems from something unexpected, but I’m not sure it understands how to create funny. That’s still the author’s role.
So let’s see what it does with emotions.
ChatGPT: There are many possible backstories that could lead to a hero in a historical romance novel having an emotional wound of not trusting people or asking for help when he needs it. Here are a few ideas to consider:
Childhood betrayal: The hero was betrayed by someone close to him in childhood, such as a parent or close friend, which left him feeling unable to trust others and hesitant to rely on anyone.
Abusive family: The hero grew up in an abusive household, where asking for help or showing vulnerability was punished, leading him to become self-reliant and distrustful of others.
Betrayal in war: The hero fought in a war and experienced betrayal by someone he trusted, leading to a deep-seated fear of relying on others and a reluctance to ask for help.
Past romantic betrayal: The hero was betrayed by a former love interest, such as being cheated on or left for someone else, leading him to become guarded and hesitant to trust anyone in a romantic context.
Previous profession: The hero worked in a profession where he had to rely solely on himself to survive, such as being a lone wolf bandit, bounty hunter, or pirate, leading him to become fiercely independent and unable to trust others.Regardless of the backstory you choose, it’s important to show how the hero’s past experiences have shaped his behavior and attitudes towards others. By understanding the root cause of his emotional wound, you can create a more nuanced and sympathetic character that readers will root for as he learns to open up and trust others again.
Okay, let’s dive into one of those. ChatGPT will “remember” about 3,000 words of previous conversation.
ChatGPT: If you want to give your hero a backstory as a bounty hunter to explain his emotional wound of not trusting people or asking for help, you could try the following:
Traumatic event: The hero became a bounty hunter after a traumatic event that left him feeling alone and betrayed. For example, maybe his family was killed by bandits or his partner betrayed him during a mission.
Self-reliance: As a bounty hunter, the hero had to rely solely on himself to survive, which led to him becoming fiercely independent and reluctant to trust others. He learned that no one would look out for him except himself and that he couldn’t afford to show weakness.
Betrayal: At some point in his career, the hero may have been betrayed by someone he trusted, such as a partner, employer, or informant. This would have reinforced his belief that he couldn’t trust anyone but himself.
Danger and violence: The hero’s work as a bounty hunter would have exposed him to danger and violence, which could have made him feel isolated and hardened. He may have experienced close calls or come close to losing his life, leading to a fear of vulnerability and a belief that he can’t rely on anyone but himself.
Sense of purpose: Despite the emotional toll of his work, the hero may have found a sense of purpose and fulfillment in bringing criminals to justice. He may have felt like he was making a difference and serving a greater good, which could have helped him cope with his emotional wounds.By using these elements in your hero’s backstory, you can create a character with a rich and complex history that helps explain his emotional wound of not trusting people or asking for help. As the story progresses, you can show how he learns to overcome these wounds and open up to the heroine, allowing them to build a relationship based on trust and vulnerability.
I kept going, asking it what situations I could put my hero and heroine in that would help him learn to trust her. It gave me generally good suggestions – shared danger, she provides emotional support, working on a task together that builds mutual respect, small gestures that can demonstrate her trustworthiness, jealousy making him confront his own feelings for her, and her making a sacrifice or taking a risk that demonstrates her loyalty to him.
These are all solid. This is the first layer of how I plot with authors. But I’m not worried about ChatGPT replacing me at this point because I know the combination of imagination, emotion and style that imbues every word written by an author (or suggested by a plotting partner) is still a stronger story element than what the bot can create by mechanically ingesting and regurgitating data.
And that’s about as philosophical as I get, y’all.
I leave you with this…
Me: write me a limerick about editing romance novels
ChatGPT: There once was an editor of romance
Whose job was to give love a chance
She fixed every typo,
Made the characters ripe-o,
And left readers in a blissful trance!