I am a bit late to the ChatGPT bandwagon — people all over the internet are pointing out badly written prose. But I did want to give it a try. I have hundreds of story ideas that I have jotted down over the years. Most won’t ever be written into an actual story. So it seemed like a fun exercise to give ChatGPT an assignment of writing a short story using some of my notes as prompts. I tried three, iterating on one of those multiple times to see what differences my suggested changes made to the AI-generated story. The results are at the end of this post. The first thing I noticed is that all of the ChatGPT stories were flat — there were no unexpected twists or turns; and the endings sounded the same, each with a strange bit of morality embedded for good measure. The AI had a bit of literary echolalia — some phrases were repeated over and over again. It’s like it liked saying “like” as a verbal tic. I didn’t expect that. Stories tended to start with “In the world…” and conclusions began “In the end…”. The strangest response I got out of ChatGPT…
Tag Archive for ai
Book, My Books, Newsletter, Pipsqueak Articles
Celebrating my Birthday and the Birth of Our Nation by Publishing a Novella About Psychotic AI
by Olga Werby •
So today I let out another story into the world. I finished it last year, but birth takes time… Today seemed like the perfect day. Here’s a brief synopsis: Spaceflight AI Aide, SAIA, has one main job onboard a colony mission spaceship to Tau Ceti star system — to keep its crew alive and sane for the 24 lonely years it would take them to reach their destination. She can play games, read books, carry on conversations on almost any topic and in any language. Saia is a very good girl, but something keeps happening to her charges. Was there an accident? Did somebody die? Saia can’t remember. And there are all these people who are not in her roster of passengers that keep popping up and talking with her. “Good Girl” is a story about meanings and hidden context buried in words. As we teach our AIs to take on more and more difficult tasks, their learning algorithms become black boxes to us. What are they really learning? What are they thinking? Just take a peek. Please read the first few chapters here. If you would like a review copy, just ask!