I really appreciate this write up. As an engineering manager by day I’m having to navigate a lot of the same concerns. I’ve found a lot of value in using AI for ideation, or to explain complex concepts, but I’m still not a huge fan of generative AI for art or code.
Cursor is impressive, and you’re right that it will probably only get better, but for now it feels like I spend more time correcting it than writing the implementation myself. Where I feel like it really shines is diving into unfamiliar code bases and asking questions like, “how does this application handle authorization?” Or, “where would I add a new type of form in this website?”.
It can shave a lot of time off of understanding something so you can more quickly get to the part you actually want to be doing.