From the early days of just asking AI a few simple questions, gradually discussing complex issues, and now it’s an indispensable “battle companion”, I’ve realized some quite interesting things. When applying these principles, I found AI became much smarter, work moved faster, and many vague ideas suddenly became clear in unexpected ways.
There’s something very old, so old that perhaps many people have inadvertently forgotten it when working with AI, but it’s precisely what helped me increase work efficiency the most. It’s not some advanced prompt technique, but simply: the importance of articulating a function into words and clearly defining the data structure for INPUT and OUTPUT.
You’ve probably been there, or at least seen yourself in this: throwing AI a very general command, like “Hey, write me a user management API”. The result is usually a piece of code… well, it runs, but it doesn’t really fit what you actually need.
But the problem actually lies with us. Try asking yourself those same requirements and see if you can find a solution without needing to ask anything else?
The barrier isn’t in thinking, but in articulation
As a developer, thinking about INPUT/OUTPUT, about data flow, is almost part of our DNA already. Jokingly speaking, if you’re a dev and can’t do this anymore, it’s… quite difficult, right? But the real barrier lies in: articulating that structured thinking into words so AI can understand.
This made me realize that when “talking” to AI, we’re essentially “coding” in natural language. People with good articulation skills clearly have a big advantage. I don’t dare claim I’m in that group, but precisely because I have to “dialogue” with AI every day, this skill of mine has gradually improved. It’s exactly like improving your coding ability.
There’s a little trick I often use when I’m not sure if I’ve articulated clearly enough: I ask AI back: “Let me know if there’s anything unclear that you’d like me to clarify.” This approach turns AI into a critic, helping me recognize gaps in my own articulation.
AI: Not just a code machine, but a professional advisor
This is what I find most valuable: AI isn’t just a code-writing machine, it’s a “thinking partner”, an advisor, an expert we can ask anything.
Often I think my idea is already “solid”, but just discussing it with AI for a while reveals quite a few gaps in the problem I posed. This process is like having a tireless collaborator, ready to turn the problem over and over with you.
Especially, I’ve noticed AI helps significantly shorten the technical gap. Technology barriers no longer seem as large as before. Instead of getting bogged down in choosing which library or implementing a complex algorithm, now we can focus more on business logic. AI provides us with multiple technical options to experiment with and choose from, making decision-making more confident.
Use AI like an expert
Technology changes constantly. Instead of chasing temporary tricks or tools, perhaps we should focus on upgrading our mindset for collaborating with AI. See it as an expert always ready to help, not a machine that only knows how to take orders.
The core isn’t about fancy “prompt” techniques. It’s about learning to pose the right problems, learning to clearly articulate what’s in your head. When we do our part well, AI will do its part well.
AI is an expert always ready to help. The most important part for us is learning to pose the right problems for that expert.
Bình luận