ChatGPT is powerful, but most people don’t get great results.
Why? Because they make avoidable mistakes when writing prompts.
I’ve seen these same...
For further actions, you may consider blocking this person and/or reporting abuse
I seriously worry about the world if “Write a 700-word blog on personal branding for developers. Use short sentences, 3 real-world examples, and end with a call-to-action.” is enough to produce content consumed by other people.
I completely get your concern. As someone who works with AI and digital transformation every day, I see this tension too. AI can produce content quickly, but it should not replace context, insight, or the human touch.
It's just that there's no point, is there? I could chat with ChatGPT myself; other people getting output with basically no insightful input is slop. Reguritated content (for now at least) that doesn't give me a human story, it doesn't provide me insight and it doesn't move knowledge forward.
I absolutely agree with that. It's horrific, but real...
The first thing I do with any GenAI tool is set up instructions! I genuinely cannot wrap my head around how anyone just... runs Copilot or even ChatGPT raw, like some kind of daredevil coder.
I spent weeks writing and tweaking these hyper-detailed scenarios for my blog posts, then GPT-5 came along and straight-up lawn-mowered them. 😆 So, now I’m redoing the whole thing — still duct-taping it together as I go. And that’s just the project-specific pile stacked on top of the system ones, times however many active projects I’ve got right now (twelve-ish, depending on how you count the ones I pretend are “paused”).
Haha, I totally get this! 😊 I’ve seen the same thing—without clear instructions, AI is like giving a supercar to someone who’s never driven. Setting up detailed prompts is half the battle.
GPT-5 can feel like it just steamrolls your carefully built frameworks, but iterating and refining is where the magic happens. Twelve projects at once? Respect. That’s real multi-prompt mastery!
Tbf, I usually only have a third of that going at any one time, but managing the edits is a special kind of learned skill 😉😆
Prompts matter! 🎯
@suvrajeet Yes, prompts definitely matter.
Interesting. I never thought of constrains as a form or indicator of creativity. By the way, out of curiosity, I tested the art of prompting a few months ago. I asked someone to write 5 free prompts and 5 with constraints using ChatGPT (so 5 pairs). Maybe I'm just completely stupid, but I was forced to guess for all 5 pairs and got 4 wrong. I didn't see any difference in quality between the free prompt and the engineering prompt.
And it doesn't even know what the word count is. If we asked for 500 words, he gave 400, if 600, then 800.
However, I agree with points 2 and 4. Patient, iterative use and providing examples lead to absolutely better results.
@daemonic01
Hey Dominik, I also agree with your perspective, this is exactly why prompting is part art, part science. I’ve found that patience and experimenting with phrasing usually make the biggest difference—sometimes a small tweak changes everything.
Thanks for sharing. Your post is helpful and easy to understand. I'll try it out!
Glad it was helpful! Excited to hear how it works out for you.
Good!
Thank You.
الذكاء الاصطناعي هو مارد هذا العصر
عليك توجيهه جيدا فربما طلباتك بدون تفكير وتخطيط مسبق تصبح لعنة وليس مساعدة
AI is incredibly powerful, but it’s not magic. Without careful planning and thoughtful prompts, it can easily create more problems than solutions. LLMs' models often show illusions in the output, and it becomes difficult to remove illusions in the outcome.