Hello friends,
From my perspective, programming with AI has completely changed for us.
The tasks that arise are:
How to keep track
You build so much functionality in such a short time that it’s hard to remember everything – documentation with full-text search is extremely important.
Need to simplify
Constantly questioning libraries and included dependencies, and testing whether it isn’t simpler and faster to just create a function with AI.
5 years ago … so it’s been a while:
“At that time, I created functions, drew many flowcharts, thought about how to connect everything, built classes, read documentation, developed security, built the platform, integrated frameworks for debugging, made translation attempts with the Windows API, worked a lot with CodeIgniter and Laravel, read books – yes, it took a lot of effort and time before I could run the first test.”
Source: FiveTechSupport Forum, post #281445
That’s true – and you have great advantages if you understand things.
But today I think:
You start with the description – verbally from the problem – and ask the AI for a solution. Then you continue step by step.
And you also need an organizational tool for the source code, because unfortunately, for “us normal developers,” it’s still not possible to always send entire projects to the AI – due to token limitations.
When debugging and further developing,
you should always proceed step by step. AI tends to often restructure the entire source, and then it becomes difficult for us to follow the new approach.
Best regards,
Otto