My biggest concern here would be false confidence in LLM code. I don't want someone to vibe code an application that takes in my password or PII if they don't k

My biggest concern here would be false confidence in LLM code. I don't want someone to vibe code an application that takes in my password or PII if they don't know how to properly audit that code. I'm sure we've all seen the posts online about how "software engineering is dead, look what I did with Claude Code by myself", but I guarantee the serial-solopenuer that's posting that has at least a few security flaws in their application that will do harm to their customers. And a trained dev can let those through as well, but I really doubt the rate is anywhere close to the same. I have coworkers that trust their LLM code _way_ too much. It does some great stuff, but I always have to do at least a little bit of cleanup or fixing. The amount of trash AI code I've had to review is kind of upsetting.