Tag: AI-assisted writing

We have a new baby!

So the reason I haven’t posted anything here in a little over a week is because we had a new baby! This one is our second, and he’s our first baby boy.

Unlike the last time, when Mrs. Vasicek went into labor around 7:30 PM and the baby was born in the early hours of the morning, she was scheduled for an induction first thing at 6:00 AM and the baby was born in the early afternoon. MUCH better way of doing things! He actually came out with the first push—the only reason he wasn’t born earlier was because the doctor was stuck in traffic, so we had to wait to start the active part of labor!

This was also our first time having a baby not in the middle of the pandemic, so we got to see the faces of all of our nurses and the doctor too, which was nice (last time, the doctor was practically in a hazmat suit with a face shield). They all got a kick out of the April Fool’s joke we played in 2020, convincing our extended family that we were going to name our daughter “Corona.” Everything went well, and Mrs. Vasicek and Little Buddy were released from the hospital the next day, which was nice.

So our lives have gotten quite a bit crazier, caring for this newborn and adjusting to all of the changes that come with that. My in-laws have been a huge help in making us food and watching our nearly 3 year-old daughter during this time, which has been really great. Still, any semblance of a routine has been shot to pieces, and it’s probably going to take at least a few weeks to recover from that.

Normally, my writing would take a huge hit from this… but I’ve actually been writing more than ever, using ChatGPT and other AI tools to experiment with writing AI-assisted stories. In fact, while I was in the hospital with Mrs. Vasicek, I actually wrote a 6k word fantasy short story during the slower moments. It was kind of funny, because just before Mrs. Vasicek got an epidural, I laughed at something funny that ChatGPT came up with, and she got really mad because there was someone in the room having joy while she was in so much pain. Then she got an epidural, and all was right with the world again.

As of right now, I’ve written five stories in conjunction with ChatGPT. The AI tools are still pretty clumsy at writing fiction, and they’re not masterpieces by any means, but they aren’t terrible either—and it actually takes quite a bit of work to make them not terrible, so it’s not like the AI is doing all of the work. One of these days, maybe I’ll do a blog post sharing some of my methods for writing AI-assisted stories.

I’m also working on the second and third rounds of revisions for Children of the Starry Sea, and it’s coming along nicely. It’s really good that I was able to finish the rough draft and first revision draft before Little Buddy was born, because I would definitely be swamped right now if that wasn’t the case. But these revisions are much easier to do, even in a sleep-deprived state: the second revision draft is all about smoothing out the flow of the narrative and making sure there aren’t any awkward transitions or loose ends, and the final revision draft is about polishing the writing on a sentence level and cutting about 10% of the words. Neither requires a ton of brain space.

So that’s what we’ve been up to out here. It’s going to take a while for things to return to normal, but we’re doing well, the baby is happy, the mom is tired but doing okay, and I’m just happy to be a dad two times over!

I wrote a story with ChatGPT…

…about a struggling magazine editor who has been swamped with poorly written AI-assisted stories by writers using ChatGPT. If that sounds hilariously meta, here is ChatGPT’s description of the story:

“Quantum Worlds” is a science fiction story about Robert, the editor-in-chief of a struggling science fiction magazine. In a last ditch effort to save the magazine, he and his wife Sarah come up with the idea of using AI to help with the submissions. The idea is met with skepticism, but it proves to be a game changer. As the magazine becomes successful again, Robert and Sarah face challenges from the science fiction community and their own daughter, who has ideas of her own about how to use AI in storytelling.

The story explores the tension between tradition and innovation in the science fiction community, and the ethical considerations of using AI in creative endeavors. It also touches on themes of family, sacrifice, and the importance of supporting the next generation of creators.

I actually think it turned out pretty well. It’s about 4,500 words, and instead of trying to rewrite it in my own voice, I’m going to send it out as it currently stands. I really hope someone picks it up!

I’d say that ChatGPT did about 40% of the work. I started by explaining the initial idea, and then we went back and forth about it for a while. When I prompted ChatGPT to come up with a twist, I had to prune the first two or three ideas it gave me, then work with it to refine it into something that would actually work (and also was a genuine twist, not just a cliche that any science fiction reader would see coming a mile away).

When it came to the actual writing, I would usually start by saying something like “write the next scene, where blah blah blah,” and then it would spit out between 400 and 600 words. I would typically have it generate a couple of responses, and I would choose the best one to copy and paste into the manuscript. A few times, I copy and pasted snippets from all of the responses to form something better.

There were a few loose ends and disjointed parts that I had to fix on my own. When I tried to copy and paste back into ChatGPT to get it to do the revisions, it just gave me mushy, unusable summarizations of what it had already written. Also, ChatGPT uses straight quotes, so I had to change all of them into curly quotes, which was annoying. But the revisions weren’t too difficult, and whenever I needed more than a paragraph or two of new material, I would copy the old material into ChatGPT and ask it to rewrite that section, giving me what I needed.

There were a couple of times when it tried to slip in “diversity and inclusivity” stuff, but overall it wasn’t all that woke or insufferable. I really do think the developers have added a filter on the backend that goes something like “make sure that your reponses have enough diversity and inclusivity,” or something stupid like that. Wasn’t hard to work around, especially for a story that isn’t political in the slightest.

Proofreading was super easy. The only things that spellcheck picked up were parts that I’d added, or words that weren’t in spellcheck’s dictionary but should have been.

It is a kind of cheesy story, though. ChatGPT was trained mostly on internet content, which is also probably why I could only get it to come up with good stuff in 400-600 word chunks. It would be interesting to use a language learning model that was trained on, say, Project Gutenberg, or the Baen Free Library. But for the story I wanted to write, cheesy is actually good.

Is there anything else you would like to tell our readers, ChatGPT?

I just want to remind readers that storytelling is a powerful tool for exploring complex themes and ideas. The use of AI and its impact on creativity and the arts is just one example of how technology is changing the world around us. I hope this story inspires readers to consider the implications of these changes and to continue exploring the intersection of technology and the arts. As always, keep reading and keep learning!