Now for those who are curious, i'm done with my first experiment and i am planning on starting a full redesign of that project, this time providing the AI with

Now for those who are curious, i'm done with my first experiment and i am planning on starting a full redesign of that project, this time providing the AI with a lot more guidance on software best practices. I'll decompose more, make it write more unit tests, etc. And make sure to keep everything in check regularly regarding code quality (still writing nothing myself though). In fact, i'll be doing exactly what i would do with a junior/midlevel developer. Except AI work probably 100x times faster. My first iteration of that project lasted about 3 days from start to finish, reaching a point where i left so much cruft accumulating that it couldn't refactor it properly without breaking absolutely everything in an impossible to recover manner... But before you say anything, i've personally seen the same thing happen at least 5 times in my career, with professional humans developing the code.