Posted for critical analysis.

Are the models correct, or biased? How does that reflect their training data and how people use them?

I would be surprised if they're substantially wrong, because of the barrels of ink that have been poured out on the topic so they surely have ample training data.

What makes those discussions tired?

The discussions are tired because people have been having them for ages without meaningful progress.

Interesting, that's exactly what TERFs say

Sure (broken clock), but they're too busy policing women's bodies to do much to actually help women. Even if trans women were significantly more likely to assault women in the bathroom, the harm done to cis women in the process of enforcing a trans woman ban dwarfs any potential benefit by orders of magnitude. The discussion only serves to divide and distract people from areas where we can actually make a difference.