AI - PART II

Okay, let’s back up a minute. I made a comment last post about AI being unreliable, but what has two thumbs and made an unreliable statement? This guy. I mentioned ChatGPT but what I meant to say was Copilot. I had written most of the first draft of the first book of a new series and thought, hey, let’s see what this new feature in Microsoft Word can do. So, I fired up the side panel and asked it for a summary, which was decent enough but got a few things wrong. Just for fun, since I still had several chapters left to write, I asked it for possible story endings. It gave me three or so, which were decent enough as far as being in-line with the story, but also rote, like these are lazy endings. I used none of them, not that I would have anyway since I already knew how I wanted the book to end. What percentage did Copilot get right? I’d say probably 70%. It overly emphasized minor characters and misunderstood major ones, and also didn’t get some of the supplemental storylines and characters. Given where we are with AI, I guess it wasn’t horrible. The book has a lot of chapters and is all third-person narration, making it complex for me to keep track of on my own. Was the AI output useful? Not really. Maybe when I go back and ask for an outline or a marketing summary. Then again, I had a much better experience last week with the first draft of another book and a different AI. Next time we’ll talk Grok.