"AI Appropriate Use?" Topic

I see a lot of opinions that are very negative about AI use in general, but I think that there is room for nuance. I don't want to read a book written by AI using a prompt. But there are some ways where it's going to be generally helpful. Before I give some examples, here are my questions:

1. Are these examples legit or would they turn you off from engaging with someone's work?

2. Do these examples require a "Made with AI" label, no label, or something in between.

Personally, I think that AI will be so ubiquitous in the future that everyone will just assume that it was used in these ways, and a "Made with AI" label will seem as silly as "Made with Microsoft Word" or "Made with digital art tools."

So, here are some examples:

1. Breaking writer's block #1: "Give me 5 random writing prompts to start my day."

2. Breaking writer's block #2: "I have three ways I can take this character, Give me a table with the pros, cons, and gotchas of each option."

3. Naming #1: "I haven't named any of my chapters. Go through my book and suggest 3 potential names for each chapter."

4. Naming #2: "Go through each of my character's biographies and make sure their name is appropriate for the time and culture they come from. Flag any that are wrong."

5. Naming #3: "Go through each of my character's biographies and make sure their name is appropriate for the time and culture they come from. Flag any that are wrong and suggest 3 better options for any that were flagged."

6. Wargame rule writing #1, "These rules are intended for WWII simulation. Give me some criticism from the perspective of a historian."

7. Wargame rule writing #2, "Update my rules to use D10s instead of D6s for a wider range of potential values. Keep the probabilities the same, but update the values in the tables for the D10 roll."

8. Wargame rule writing #3, "My chance card titles don't feel flavorful enough. I want them to feel grittier and more realistic. Suggest some better titles."

9. Wargame rule writing #4, "Before I send this to my friends to review, go through my rules and check for contradictory situations, unexpected edge cases, and situations that could arise that are not covered at all. Give me feedback as a bullet-point list summary."

10. Wargame rule writing #5, "Here's some common situations that arise in wargames: 1) Vehicle groups that look like parking lots; 2) Multiple units ganging up on a unit that's in a different (but appropriate) formation; 3) Simultaneous charges into melee that are hard to resolve. Check my rules and see if these situations could arise."

11. Wargame rule writing #6, "My combat modifiers table is way to big. Analyze the table and suggest some ways to slim it down by combining similar modifiers or eliminating inappropriate ones."

- K.

Yes. I do lots of that sort f thing. It is talking to myself.

As for your initial point, I will keep singing the same song: all machines are amplifiers. If you are lazy, uncurious, uninspired and uninformed, you could write a book without AI and it would be terrible but at least it would take you a few months. Now you can churn out that terrible novel in a week. But if you are passionate, if you have delved into your topic and thought about it, you can now write a novel faster and better, and for the reasons you list.

I'm using it for apps and game rules right now, and WOW. Claude is worth that $20 USD per month.

I also agree that it will be ubiquitous someday, but for a while we need the "made without AI" label on some things.

For #2 I'm concerned that if you don't know where the character is going you don't have anything to say and AI is not going to make your story good, it's just going to help you get it done sooner. I would say a better use would be "I want this character to figure out this, what could he encounter that would cause the epiphany?" Or "I don't want the character to seek this solution yet, what are some reasons he can't or won't?"

For #3, based on my experience, don't just tell it to name chapters. Tell it the spirit you're going for. Should they all be dramatic? Cryptic? Informative? Avoid spoilers? Tell it why you WANT chapter titles instead of just numbers. If you don't tell it where you're going, what you want from the output, it just does stuff. You'll get output, but it might not be what you want and it might be terrible.

#4 and #5 are spot on. I find tasks like that draining and they slow you down. Now you can get it done well, for almost free, instantly.

#6 I did an experiment like this recently and was startled by how well the free Claude account handled it.

#7 It can absolutely do that, and it can tell you what probabilities are for combinations of things that may occur across your resolution system. EG if the to-hit system works like the and the damage system works like this what are the odds that unit X will take out unit Y in one shot? How many shots from A can a B take before it's knocked out? Where is the point of diminishing returns when adding more bombers to this target and is that historical?

#8 Absolutely. I would make that prompt longer and give more direction about the feel you're looking for. "Gritty" is one component, but should these card titles be in the voice of the commander? or the lingo of the soldier? How long can they be? Should they talk about what's happening in the game or the big picture? Give examples of the kind of thing you're looking for.

AI LOVES example output.

#9 I have not tried this yet, but I'm looking forward to doing things like this, maybe this week.

Alongside this, we may be able to drop in the rules and say "write an example consistent with the rules that demonstrates the procedure and make sure this or that result are included." This not only produces the example for the rulebook, but it's testing the rules since it's a micro-playtest.

#10 This might or might not work. AI doesn't understand anything, it just creates the appearance of understanding. I have many times had Claude anticipate things that yes, a user would experience, but I would not count on it. It could also hallucinate here. So I would certainly try this, but you're at the edge of what works and what does not work.

#11 I suspect that would be extremely helpful.

I look forward to having more time to experiment with this stuff in a couple weeks. I'm working on some apps right now and I'm ASTONISHED at how Claude Code just magically does stuff, including writing and running tests, anticipating problems users might have, improving user experience, we're at the dawn of a very different world…

So at this point would you say #1 would invalidate the "made without AI" label? Or is that the only one that wouldn't?

I wasn't as worried about the exact prompt engineering as the gist of what is being asked for as it relates to labeling or an in/appropriate use.

For #2, you voiced a concern "that if you don't know where the character is going you don't have anything to say and AI is not going to make your story good, it's just going to help you get it done sooner." Couldn't it also help a good writer do it sooner? At that point one novel would be good and one would be bad, and they would have the same labeling – the same result as before AI.

For #4 and #5 – would you be concerned if #5 were labelled "Made with AI" but #4 wasn't?

You also wrote – "Alongside this, we may be able to drop in the rules and say "write an example consistent with the rules that demonstrates the procedure and make sure this or that result are included." – I think that it won't be long before everyone should just assume that's how all rulebook examples are generated.

Also, final proofreading is always on the human, so any uncaugut errors are human errors.

Your examples are mostly using AI to analyze or make suggestions for an existing piece of work

That is probably the sweet spot for AI right now

Brainstorming is a powerful technique – using AI to brainstorm ideas for I'm sure will become standard practice