AI

I hear there’s a bit of conversation on the Interwebs about AI and writing. I took Microsoft’s Copilot out for a couple of short test drives and here’s what I found. No, I didn’t have it write a book for me. Although, I might try that and see how it goes just for fun. I took the manuscripts of Assignment Day and the new book I’m working on and asked it how I could make the story better. For the new book, which I ran through AI first, it gave me a list of about seven bullet points, saying I should use foreshadowing, more inner monologues for the characters, and some pretty general stuff like pacing. It picked up on the fact that there are multiple characters, which seemed like a good thing. Contrast that with its output for Assignment Day. The book starts off with a chapter about a character who appears later in the book, but he’s not the narrator or main character. Copilot threw out eight items, more than half of which named that first guy as the main character, and no mention of the actual first-person narrator. The list was similar to my first experiment with suggestions on more dialogue, pacing, descriptions, and proofreading.

Next, I asked for a summary of Assignment Day. This went a little better. It missed how a scene fit into the overall plot, but overall at least Copilot recognized that character from the prologue wasn’t the main one, but it also didn’t name the actual narrator and protagonist. At least it came up with a useful summary: “The story intertwines themes of technology, virtual reality, and the quest for truth, culminating in a suspenseful and engaging adventure.”

Were the answers to my question useful? For the most part, no. Would I get better results if I got more specific with my questions and honed in on certain aspects of the books? Probably, but not right now. More to come as I experiment with AI from time to time and figure out how it can actually help.