This is perhaps where our perspectives differ, because I see the usage of LLMs not as an external third-party (another team per your example), but instead as an extension of one's self. Given that lens, I'm highly sensitive to the quality and function of its output, because ultimately its contribution is my responsibility.
replyI appreciate not everyone feels this way, but that's why I personally would be anathema not to read its code.
My philosophy is just to Duck-type the program: "If it walks like a duck and it quacks like a duck, then it must be a duck"
replyI don't care if the duck is wet spaghetti inside, it does what I need it to do within the parameters I can measure.
If it fails to quack or walk later on, I have production alerts for that and I'll deal with it then.