Post #2089829

One of things that many totally overlook is that reviewing AI generated code is harder. It doesn't look like regular code. It has mistakes in places that people wouldn't make. It doesn't have appropriate abstractions. You can't let your guard down like you can with a human engineer. It's like "somewhere in this massive pile of code, there will be a mistake that not even a junior engineer would make. But no clue where!" To be fair, I'm sure people can use GenAI to write decent code. But that requires a lot more discretion than most people are doing.