It's hard to say definitely whether, as PC World Does, this marks a kind of first for autonomous vehicle accidents in San Francisco because human error could be the root cause. Basically, it's confusing, and that's the point, as the Wall Street Journal's Digits blog also argues.
This all happened a couple of weeks ago on January 8th: A Nissan Leaf hit a parked Prius. Now, with so many Priuses in town, smashing into one would seem to be the San Francisco equivalent of running into a tree or a telephone pole. But the notable item here is that the Leaf was a test car operated by Cruise Automation, and had just been in autonomous mode. It had finished switching to human driver mode before impact, though how long before remains unclear. There were no injuries and both cars suffered minor damage, nothing more.
The crash took place around 7th and Bryant, when "The vehicle began moving in its lane to the left, then began correcting to the right at which point the driver decided the take over manual control,” the company said in a report to the California Department of Motor Vehicles. “After taking over manual control, the operator did not change the path of the vehicle and it collided with an unoccupied Toyota Prius."
We know of the incident because the California DMV requires there to be a report of every accident that involves autonomous driving tests on public roads. Most of those accidents have occurred in Google's backyard of Mountain View, such as one fender-bender SFist noted last summer, reported as the company's first.
According to Cruise Automation's CEO “There was enough time [the driver] took over manual control but unfortunately made a mistake.”
In the end, the incident highlights the grey area between an automated driver and a human one, and feels important because that's how it looks like driverless cars will start out — right in that zone.