@smcgrath.phd That study is a group of researchers training their own AIs that have the same category as the ones used in hospital. They did not test models "used in hospitals". Their badly trained models but did a bad job. They basically straw manned the AI systems by making bad copies.

I think the AI is the least concerning bit of this story. One side of a conflict using a "news" site trying to discredit academics or experts happened well before AI slop. AI slop is definitely a problem, but it is not the cause of the problems in this article.