I really wonder do they even audit the discussion outside their own comments. You can absolute notice how popular the option of not "auditing" code is in their AI-tribe. Best they do is have it review it or make tests... Not actually read the later either...

And then when someone does audit the stuff output is questionable... Telling that it is most likely not that brilliant.

I am thinking that just maybe the "load bearing" (for AI terms) components should have quite more oversight. Which might go away unless they are actually human authored...