LLMs keep learning, growing, and improving.
But they are only as good as the scenarios that they are dropped into.
For instance;
A new code library that doesn’t log errors- great when it works, not great when it fails.
Requirements that use “generic” nomenclature and not what everyone else uses, only good if you’re a senior dev.
QA Test Cases that generate failure scenarios for tests which don’t apply to your stack? Add extra work.
Diagrams that point to systems that don’t exist (but should), don’t give you the current state.
If you’re not cleaning up and working with your AI skills, models, routines, etc.
You’re just creating extra cleanup work for people to do in their daily work.