How many autonomous cars are there compared to human-operated? Far far more.

I think you meant less.

Ideally, you'd be correct and we should be looking at per capita incidents- like how many incidents per 100 miles on the road or something. But the article just cited a flat number of incidents without contextualizing, which as you've pointed out can be misleading. Without knowing the ratio of AVs to human-driven vehicles, the best rebuttal that could be offered is "Yeah, but how do those 90 incidents compare to how people drive?"