Is "working" enough?
Code or configurations created by AI are not necessarily "correct" even if they work.
Paste a part of the code or configuration you are concerned about.Who is supposed to see these logs?
For logs and metrics, valid design is defined by the "premise of who watches", not just making them.
Paste a few lines of actual logs.When it breaks, where does your responsibility end?
Systems will inevitably break.
The issue is the boundary at that time.
Can you explain the boundaries AI created?
AI is good at creating boundaries.
But whether you can explain them is another matter.
Is it okay to "decide not to decide" now?
Not deciding is also a design choice.
However, the responsibility remains.