You do not explain why these rules are stupid, much less "the stupidest".
Both rules, on first sight, make __a lot__ of sense.
Both rules, on first sight, make __a lot__ of sense.
A transparent but poor decision is not really much better than an opaque poor decision.
There's no magic objective moralism box that you can put a decision into and have it weigh its consequences and a little green or red light turns on.
Good for whom? Under what assumptions was it made? In what context does it hold?
All of that is transparency.
Creating mountains of documentations on the model for a regulator will not do that.
Similarly, using the input/training data quality as vehicle to avoid, e.g., discrimination is weaker than directly targeting an outcome.
That could be a reasonable decision for people to make. If observability comes with accountability then it might eventually end up as the same thing.
But it is better, or rather more desirable, as it is easier to fix and testable against rules which might be very important in certain contexts.
Forbidding AI to identify race is forbidding AI tools to help serve those people better or even detect that they are being served poorly.
It assumes even that targeted advertising by race or gender orientation is only exploitative and not seeking to help fill key needs.
AI is only made by racists and religious jobs as a basic assumption seems like a very stupid assumption.