But I would ask the people living there what they perceive, not just relying on a news reporter.
Cons: - I'm a cyclist, and I often make eye contact with drivers to ensure they're aware of me. Without a driver, there's not a good way to ensure the car knows I'm there. That being said, I've personally never had a close call with one on my bike. - As mentioned in the article, they can get in the way of first responders. I don't think that's justifiable and should be something that these companies prioritize before expanding their operating hours and range. So yeah, a couple anecdotes and thoughts from someone in the area. They're not perfect, but I think the upside potential is great and the city should be working to accommodate them and get human drivers off the roads as much as possible.
However I also don't want them sideswiping a cyclist in a rush to get out of the way of an ambulance.
Autonomous agents -- humans or otherwise -- base their decision on the imperfect information they have. No one has a global, perfect view of everything, and so the perception of how well other agents fulfill their promises (formally defined as intentions made known to an audience) will always be based upon local, imperfect information.
I can mention other frameworks -- Cynefine, and the error where one confuses a Complicated domain (that can still be accurately modeled) with a Complex domain (that is impossible to accurately model). Or what James C Scott discusses an idea called "legibility" and the fallacy in imposing legibility on complex systems in his book, Seeing Like a State.
Perceptions matter. Blaming the participants of a system for their being uninformed will not lead to voluntary cooperation, much less reliable systems that involve both machines and humans.
There's going to have to be (or there may already be) a rule against accusing people of being ChatGPT, but I can't believe that this is an argument that the "imperfect information" that average people have about some condition or event is somehow more important than the actuality of the condition or event precisely because of how wrong average people can be?
Because voluntary cooperation? I should only be concerned with that if I'm doing PR work for these companies. It's their job to sell safety, the only thing I'm concerned with is when people are lying. Or intentionally confusing the public about some fact, polling the confused public about what they think the facts are, then reporting the poll to further confuse the fact in lieu of simply reporting the data.
Now (literally yesterday) a driverless Cruise car trying to make a left turn was yielding to me walking across the intersection, and then suddenly decided to go before I'd cleared the intersection. Would it have hit me if I hadn't scurried out of the way? I don't know, but it didn't inspire confidence, and it doesn't take many experiences like this to turn public opinion. Cruise in particular seems to have made their cars more aggressive in my small personal sample size.