Why am I not producing AI slop?

There's lots of concern and discussion about the quality of AI models writing code. I am writing fairly ambitious code in complex codebases and I have really been seeing great results with Claude, OpenAI, gemini, etc. I don't have any particular process or magic prompts, but I have been developing software for 30 years and I use AI as if they were a developer I manage. I try to be clear on the goal of the code, context that should help, and then I don't too picky about exactly how the code should be written. I test it and give feedback with how the code should work. Why are so many people complaining about AI slop? Is it really the shortcomings of the AI models or are they using AI wrong? There appears to be entrenchment going about whether there is AI hype or not. I'm wondering it's really almost a personal issue, like it depends on how you work with people. I generally tell myself that if I'm not getting the results I want then I need to improve how I am using the AI. And I've gotten great results. Am I alone on this?