The company left out some key details regarding the incident involving one of its robotaxis and a pedestrian.
On October 2, 2023, a woman was run over and pinned to the ground by a Cruise robotaxi. Given the recent string of very public malfunctions the robotaxis have been experiencing in San Francisco, it was only a matter of time until a pedestrian was hurt by the self-driving cars. New reports, though, suggest that Cruise held back one of the most horrifying pieces of information: that the woman was dragged 20 feet by the robotaxi after being pushed into its path.
The LA Times reports:
A car with a human behind the wheel hit a woman who was crossing the street against a red light at the intersection of 5th and Market Streets. The pedestrian slid over the hood and into the path of a Cruise robotaxi, with no human driver. She was pinned under the car, and was taken to a hospital.
But this is what Cruise left out:
What Cruise did not say, and what the DMV revealed Tuesday, is that after sitting still for an unspecified period of time, the robotaxi began moving forward at about 7 mph, dragging the woman with it for 20 feet.
read more: https://jalopnik.com/woman-hit-by-cruise-robotaxi-was-dragged-20-feet-1850963884
archive link: https://archive.ph/8ENHu
The question isn’t whether they’re infallible, just whether they’re less fallible then humans, which is a far lower bar when it comes to driving.
A bold argument to make on a thread about a car dragging a woman stuck under it for 20 feet and then the company covering it up. The one story contains both the technical problems that people like me have been warning about since day one (bad at edge cases) and the structural and political problem of corporate control over infrastructure.
Yeah but you don’t have to look far to see humans doing way worse with cars. Even in this case the most reckless, irresponsible actor wasn’t the AV, or the company but the person who did the initial hit and run in the first place.
Ideally we’d get all cars off the streets, there use is dangerous in and of itself. But after being around these things for 3 years now I’d take them over the human drivers who I repeatedly see speeding through intersections.
As far as I know, there doesn’t exist a single shred of actual, empirical evidence that we can make self driving cars actually better than humans, outside of our faith in technological improvement. Companies used to publish their data on simulated injury rates for their internal testing, and they were way way way worse than professional human drivers, like taxi drivers, which is what Cruise is trying to replace.
I don’t want to be right. I want self-driving cars that work. It’d be personally very convenient for me, so much so that I would buy a brand new car for the first time in my life (I’m in my mid thirties) if I had actual, robust, empirical reason to believe that they work as advertised.
That article was written over a year ago and since then a lot has changed. Both waymo and cruise have now been approved for av taxis in San Francisco. This decision seems to not have been motivated by hype but a good track record with less incidents then a human driver. Cruise claims it’s greater than 50% better then a comparable human driver and while that may be just their own flawed study this article also agrees that they’re about even if not better then human drivers.
You’ll probably have to wait a while to buy one though as these are decked out with a larger assortment of sensors compared to a Tesla and probably cost a couple hundred thousand all said and done. They are also specifically trained on the surface streets of San Francisco, so they probably won’t be able to take you on your commute any time soon. Hell id give it another 5 years before their even able to take you to the airport 10 miles south of the city since it would have to get on the freeway. But in this relatively limited problem space they do quite well.
Both of those links rely on the same self-reported data by Cruise… We’ll see, I suppose! Happy to be proven wrong. I live on a frozen hilly dirt road, so realistically, for me, it’s not going to happen.
The second one relies on it for the number of incidents, which cruise is required to report to the police and in my opinion can be taken as fact. You can try to hide who was at fault for the incident, like they’re doing in the original story, but it’s very unlikely you’ll get away with hiding that an incident happened. If cruise doesn’t report it then the driver or pedestrian that got hit sure as shit will, and they’ve got their branding all over it so they know who to report to. If they report it and cruise gets caught trying to cover it up, they’ll be in deep shit, enough to not make it worth trying to do so. They just lost their license for omitting some footage after a report, imagine what would’ve happened if they didn’t report at all.
Sorta… the second does some pretty thin analysis, and the refers right back to the same report, of which i am skeptical considering they were just caught bending the truth in their favor.
We need a robust and publicly accountable third party investigation. I fear we don’t have the political and institutional infrastructure to do that, and it’s going to be like how even just 50 years ago the FDA regularly was letting dangerous shit fly that they absolutely should’ve caught and that companies were straight up lying about.