Yeah, I think the article writer is too confident in the idea that these issues won't be solved:

"Those defending the poor performance of vibe coded apps have parroted the famous lines AI guys used in the past - “It is only days old” and “this is the worst it gets”. They make the mistake of equating human app development time with generative AI development time (hundreds of times faster). A vibe coded app is, in human hours, not only a few days old but closer to two or three years old. If a brush engine or transform tool works badly after the human equivalent of three years of development that is a demonstration of bad planning or an AI model not doing a good job."

I don't understand how this argument makes sense. Sure, yes, the vibecoded Photoshop clone is three human years old; I agree. It would probably take me about three years to build something like this. But that doesn't change the fact that its development velocity is, and will remain, "three human years in a few LLM days."

So if the brushes are a pain point, tell the LLM to focus on brush performance and let it run for a few hours.

Well I did it with actual research, and the goal was "make something that works"