Anonymous #1 3w

It just can't seem to handle fine nuanced details outside of the training data. I am trying to get it to do complicated engineering calculations to be fair. I run into frequent hallucinations. And when I write C or C++, it's allocation strategies are terrible and unmaintainable. The one thing I've got it to perform well at was writing a basic fast API service. It got the scaffolding up pretty quick. Or writing large amount of boilerplate.

1

Anonymous OP 3w

Are you trying to vibe code the whole program at once or piece by piece?

1

Anonymous #2 3w

Small sections within the same context with explicit instructions for each step. I usually waste more time than I save, so I've recently just taken AI out of my workflow completely other than as a stack overflow replacement. Even then, I often find just slowing down and reading the docs like I used to more helpful.

1

Anonymous OP 3w

With the app I made, I used Python 3.13. Building the foundation and making sure it works the way I need to through the command line, then building the GUI and being very specific on what I want

1