AI coding tools are getting good enough to actually ship things, which is kind of a problem for learning

Been tinkering with a SaaS side project for a few months and the gap between what I can ship now versus a year ago is genuinely strange. Not in a purely good way either. The tools are good enough that I can move fast through parts of the stack I barely understand. Which works until it doesn't, and when it breaks I'm staring at code I didn't fully write trying to debug something I can't fully reason about. That's a new kind of stuck that feels different from the old kind. What keeps nagging at me is whether people building with these tools are actually learning anything transferable or just getting faster at generating things that mostly work. For someone treating this as a hobbytoproduct pipeline the productivity gain is real. For someone trying to actually grow their skills it might be hollowing out the parts that matter. That Chinese models post from earlier this week got me thinking about this more. As these tools get cheaper and more capable the barrier to shipping keeps dropping, but the barrier to understanding what you shipped might be quietly going up. Curious whether other people building side projects have hit this wall or if the learnbydoing argument still holds when the doing is increasingly delegated.