Because, for all of the awfulness they bring to the rest of us, they are human.

Humans who the other humans desperately need to be stripped of their wealth and power, and for whom the doing of which might offer them some small chance to save themselves from the yawning void of more more moremoremoremoremoremore

It is not obvious that LLMs are ever going to be much more than they are now, and a major change in architecture is going to take a long time to get up to this level, especially as all the money chases LLMs and the video cards let the magic smoke out day by day.

I agree that AI is going to eat the world and people really do have no understanding of the ground shaking under their feet. “Tomorrow will be basically like today, but with bigger TVs,” is a strong bias to be pushing against. But that all said, that does not mean that LLMs are it. It is entirely possible that their cap is slop and no higher.

On priors, you’re talking about the Cass report, which was in fact massively flawed, and I saw a great deal of discussion on it. Among other issues, they tossed out a lot of studies, and wrote the report from the stance that the only studies worth looking at would have been double blind studies. I’m not sure if you’ve considered this specifically, but it’s pretty damn hard to blind starting to grow boobs or facial hair. Cass, the supposedly neutral party conducting the study, was handing out copies of “Irreversible Damage”, a scaremongering antitrans book about how the poor helpless young “women” were being seduced by big trans. It was garbage, and there’s plenty of writeups on it.

https://kinesismagazine.com/2025/02/16/methodological-and-ethical-failures-of-the-cass-review-why-it-falls-short-as-a-guideline-for-trans-healthcare/

https://www.scientificamerican.com/article/the-u-k-s-cass-review-badly-fails-trans-children/