During the panel discussion at @what_the_stack something around this came. The discussion was around the understanding what you are building. The example I gave was, back in the days, around 2011, I was building an feature to send an email, it was using asp net, it took me weeks to read the docs, figure out if the code I wrote work's? With AI you can experiment multiple ways to solve an problem, but understanding what you are building or what you want to build is the key. AI is making me help things ship faster, but as a engineer, its my job to realize what's useful for my users and what's not.