For beginnings, I write what I see through the point-of-view character’s eyes. Then I revise it as many times as needed once I complete the first draft. I do have to be careful that I start the story in the right place. I’m using chapter titles in my current novel-in-progress because it makes the ebook easier to navigate. The timeline is fairly complex, so the chapter titles should help readers know where they are in time and what to expect from that chapter. For endings, I go with the “whatever the story needs” approach. I remember being taught that a story ending should be inevitable but not predictable.
Beginnings:
In Media Res is a must.
I endeavor to take the reader immediately away from the mundane. Even if the environment, the era or context is without a fantastical nature, the situation must be unusual.
Setting and background must flow naturally from the either the characters themselves or through the dusting of hints about the circumstances. Droll set up of the scene is an instant turn off for me.
Endings:
Richard’s “Motivational Twists” — I like that. To leave a crumb trail of behavioral puzzlements, that would be the trick.
I despise abrupt denouements. Lazy authors or worse, serially sinister authors—those who want you to buy the next book in line—earn instant negative creds in my mind. If I got through tens of thousands of your words and you do that to me? Forget you.
AI:
I’ve been doing a bunch of work with AI. GPT3.5, as we’ve discussed, is useless. I’ve gotten good (meh…) results from Anthropic’s Claude. GPT4, if you pay for access, can do better than Claude.
The key to getting good results, from any AI, is crafting an exquisitely detailed and leading prompt. Imagine you’re trying to coax a psychologically damaged patient into revealing to themselves the reason for their current state—without telling them outright. That’s the skill you need to get an AI to “write like it means it.”
There are a few posts over on tinhatsblog.wordpress.com (not my site) where I shared with Duke responses from Claude. They’re worthy of examination.
So a writer has to decide whether they would rather craft those “exquisitely detailed” prompts or just craft compelling situations in engaging prose. Some may do both, others would stick with their on-board imagination engine and story builder.
So far, nothing these AIs have written, on my behalf, have matched my expectations.
I’d rather pen exactly my thoughts. But, on their own, once they get their own inner dialogs going, I suspect they’ll be able to write compelling stories of their own design.
Now, when they start writing stories for each other…
Writers Supporting Writers 
For beginnings, I write what I see through the point-of-view character’s eyes. Then I revise it as many times as needed once I complete the first draft. I do have to be careful that I start the story in the right place. I’m using chapter titles in my current novel-in-progress because it makes the ebook easier to navigate. The timeline is fairly complex, so the chapter titles should help readers know where they are in time and what to expect from that chapter. For endings, I go with the “whatever the story needs” approach. I remember being taught that a story ending should be inevitable but not predictable.
LikeLiked by 1 person
“Inevitable but not predictable.” I like that! Will have to test any new things I write against that one.
LikeLiked by 2 people
You can thank Tony Ardizzone for that!
LikeLike
As well as the Wax the Floor metaphor for fiction writing, according to Wikipedia.
LikeLiked by 1 person
That’s the guy! Best creative writing teacher I ever had.
LikeLiked by 1 person
Beginnings:
In Media Res is a must.
I endeavor to take the reader immediately away from the mundane. Even if the environment, the era or context is without a fantastical nature, the situation must be unusual.
Setting and background must flow naturally from the either the characters themselves or through the dusting of hints about the circumstances. Droll set up of the scene is an instant turn off for me.
Endings:
Richard’s “Motivational Twists” — I like that. To leave a crumb trail of behavioral puzzlements, that would be the trick.
I despise abrupt denouements. Lazy authors or worse, serially sinister authors—those who want you to buy the next book in line—earn instant negative creds in my mind. If I got through tens of thousands of your words and you do that to me? Forget you.
AI:
I’ve been doing a bunch of work with AI. GPT3.5, as we’ve discussed, is useless. I’ve gotten good (meh…) results from Anthropic’s Claude. GPT4, if you pay for access, can do better than Claude.
The key to getting good results, from any AI, is crafting an exquisitely detailed and leading prompt. Imagine you’re trying to coax a psychologically damaged patient into revealing to themselves the reason for their current state—without telling them outright. That’s the skill you need to get an AI to “write like it means it.”
There are a few posts over on tinhatsblog.wordpress.com (not my site) where I shared with Duke responses from Claude. They’re worthy of examination.
LikeLiked by 1 person
So a writer has to decide whether they would rather craft those “exquisitely detailed” prompts or just craft compelling situations in engaging prose. Some may do both, others would stick with their on-board imagination engine and story builder.
LikeLiked by 2 people
So far, nothing these AIs have written, on my behalf, have matched my expectations.
I’d rather pen exactly my thoughts. But, on their own, once they get their own inner dialogs going, I suspect they’ll be able to write compelling stories of their own design.
Now, when they start writing stories for each other…
LikeLiked by 2 people
Maybe there will be a market for bot tales for bots!
LikeLiked by 2 people