Personally, no, haven't seen it in professional settings. Colleagues using AI heavily did not show any superhuman development speed, but definitely increased review burden compared to writing code by hand.

I'm not seeing it if that's true. None of my vendors, none of my open source upstreams, and our own dev team haven't actually been getting software out faster than they were a couple of years ago. As a sysadmin docker had a bigger impact on my work in its first few years than LLMs have in theirs.