I like the article for its contribution of numbers. And I think overall the narrative is valid at this point in time, however I also see three problems with it that show a certain degree of unfounded pessimism. 1: it assumes that the upfront cost of these cars will stay the same, which, of course, is nonsense. There is no reason why economies of scale will not bring down that cost by half or more. 2: It uses the 1.5 people/car statement a bit to "linearly" -- only because right now there is such a ratio doesn't mean that cannot be reduced. In fact, it is not clear whether that number is not merely #car/#employees. Otherwise it would be very inconsistent with Vogt's statement about the lack of need to further optimize. 3: the last few paragraphs seem to be trying to say that these cars are not safe, and the reasons give are a few examples and that they were shut down, but it doesn't compare to the number of accidents caused by humans!
Robotaxis theoretically require no humans and can run 24/7. In practice they need fueling, cleaning, and frequent interventions by remote human drivers to get them out of jams.