Generative AI only really helps (a meaningful amount) developers where writing code is actually the bottleneck. I have not been in a such a position for years and years.
Same experience here. Today I spent 5 hours debugging code, and wondering how I could put 3 contradicting specifications inside while negotiating some weird stuff that does not concerns me, and then I spent 3 hours deleting 100 lines of code and writing 100 lines of code to please everyone. I fail to see how a LLM could have helped me, and I've been doing this for more than 10 years.