Need help related to AI article prompting.

The issue usually isn't the model, its that the prompt is asking it to "write well" instead of giving it clear constraints.

A few things that helped me, tell it to use plain, concrete language and avoid poetic or metaphorical phrases, explicitly ban filler like "imagine", "picture this", "heavy eyed", "journey", "unlock" etc. Make it explain concepts directly instead of creating atmosphere and have it write a first draft than run a second prompt that critiques and rewrites anything that sounds like AI or vague marketing copy. If you have examples of your writing use several samples instead of just one so it learns the overall pattern better.

For your example I would simply write, "This is the transition between wakefulness and sleep, known as the hypnagogic state, during this period your thoughts become less analytical and more associative making it easier to relax and process ideas, theta brain waves naturally occur during this stage as well as during deep meditation." Its much clearer and gets the point across without the unnecessary fluff.

Hopefully this helps :D
Sorry for replying late. I had something important going on in life couldn't reply to it. The AI model uses too many metaphorical phrases i will use your example to set it to achieve the desired outcome. Thanks
 
One thing that actually cut AI fluff for me was adding a negative example directly in the prompt. Paste an example of bad output and write "never write like this" right next to it, AI respond much better to concrete examples than abstract instructions like "be clear." also try ending prompt with "if you catch yourself writing a metaphor, delete it and restate as plain fact instead." That one instruction alone reduce the poetic filler significantly.
Thanks, after giving prompt about metaphor and poetic in detail, it started to use less "heavy-eyed" and other words, but i need to give him one self-written example where he could understand more about what i want.
 
I start by generating a large list of titles first. Then I remove the ones that are too similar, would produce nearly identical content, or are too generic.

After that, I ask the AI to assign each title a core topic and a writing tone. Once that's done, I give it competitor URLs and ask it to generate the article structure based on those. Finally, I generate each article separately instead of creating everything in one go.

I never generate full article in one go. I generate each part of article separately. Just so that i can check and reevaluate it myself.
Me too. I just give him the outline and URL and ask him to do it section by section so i just get to know what section feels off and what i need to do there.
 
Hey, BHW members
I need your help with ai article writing. I let my AI do the research and writing, and I gave him a detailed prompt on what to follow and not. But in the end, it still writes generic shit that has no meaning to it. I usually rewrite myself, but sometimes it takes a lot of time to rewrite in my style. I even gave him my style to replicate, but he still wrote something else. ( I use claude and chat gpt Free*)
Look at it...
(Here's what it feels like: heavy-eyed and inward, like the moment right before a dream starts, the fog right after you wake up before you've opened your eyes. Sleep scientists call this the hypnagogic state. Thoughts stop feeling sharp, and time gets a little loose...
This is the state from last section, the one where the critical faculty softens and stops arguing with everything you say. Theta shows up on its own during deep meditation, in the minutes right before sleep, and sometimes right after you wake up, before the day fully starts.)

"What's the point of heavy-eyed and inward"
How do you guys create output which avoid this type of writing?
One of those ways was by not letting the AI compose the whole article at once. Instead, I let it generate an outline and ideas first before moving on to the sections one at a time, making sure that the langiuage used was concice and straightforward by following certain rules.
 
free versions are the problem, both claude and gpt free tiers cut corners on nuance way more than people realize, paid tier alone fixes half this
 
free versions are the problem, both claude and gpt free tiers cut corners on nuance way more than people realize, paid tier alone fixes half this
Claude is kinda expensive for me right now so all i gotta do is stick with the free version and do as much as i can
 
Hey, BHW members
I need your help with ai article writing. I let my AI do the research and writing, and I gave him a detailed prompt on what to follow and not. But in the end, it still writes generic shit that has no meaning to it. I usually rewrite myself, but sometimes it takes a lot of time to rewrite in my style. I even gave him my style to replicate, but he still wrote something else. ( I use claude and chat gpt Free*)
Look at it...
(Here's what it feels like: heavy-eyed and inward, like the moment right before a dream starts, the fog right after you wake up before you've opened your eyes. Sleep scientists call this the hypnagogic state. Thoughts stop feeling sharp, and time gets a little loose...
This is the state from last section, the one where the critical faculty softens and stops arguing with everything you say. Theta shows up on its own during deep meditation, in the minutes right before sleep, and sometimes right after you wake up, before the day fully starts.)

"What's the point of heavy-eyed and inward"
How do you guys create output which avoid this type of writing?
The problem is not the length of the prompts it is the quality of the inputs. I get better responses if I feed the AI actual samples of my writing style, target audience information, and a list of things to avoid. I do not depend on AI for the first draft anymore.
 
dont ask it to “write a good article”, thats when it starts doing poetic filler. give it facts first, then make it write one section at a time in plain english. also paste a bad example and say “never write like this”, works better than just saying no fluff.
 
Your prompts are too vague be strict & specific by forcing structure, tone & examples, then edit in2 passes instead of expecting perfect output in one go
 
Hey, BHW members
I need your help with ai article writing. I let my AI do the research and writing, and I gave him a detailed prompt on what to follow and not. But in the end, it still writes generic shit that has no meaning to it. I usually rewrite myself, but sometimes it takes a lot of time to rewrite in my style. I even gave him my style to replicate, but he still wrote something else. ( I use claude and chat gpt Free*)
Look at it...
(Here's what it feels like: heavy-eyed and inward, like the moment right before a dream starts, the fog right after you wake up before you've opened your eyes. Sleep scientists call this the hypnagogic state. Thoughts stop feeling sharp, and time gets a little loose...
This is the state from last section, the one where the critical faculty softens and stops arguing with everything you say. Theta shows up on its own during deep meditation, in the minutes right before sleep, and sometimes right after you wake up, before the day fully starts.)

"What's the point of heavy-eyed and inward"
How do you guys create output which avoid this type of writing?
Instead of asking AI to write the whole article, I let it create an outline and key points first. then I write or edit each section myself. teh content feels much more natural.
 
Hey, BHW members
I need your help with ai article writing. I let my AI do the research and writing, and I gave him a detailed prompt on what to follow and not. But in the end, it still writes generic shit that has no meaning to it. I usually rewrite myself, but sometimes it takes a lot of time to rewrite in my style. I even gave him my style to replicate, but he still wrote something else. ( I use claude and chat gpt Free*)
Look at it...
(Here's what it feels like: heavy-eyed and inward, like the moment right before a dream starts, the fog right after you wake up before you've opened your eyes. Sleep scientists call this the hypnagogic state. Thoughts stop feeling sharp, and time gets a little loose...
This is the state from last section, the one where the critical faculty softens and stops arguing with everything you say. Theta shows up on its own during deep meditation, in the minutes right before sleep, and sometimes right after you wake up, before the day fully starts.)

"What's the point of heavy-eyed and inward"
How do you guys create output which avoid this type of writing?
I've run into this too. I usually tell the AI to write in a simple way and avoid trying to sound too creative.

The first draft is rarely perfect, so I just make a few edits and move on.
 
Back
Top