Who’s at Fault in Uber’s Fatal Collision?

Publication Type: 
Other Writing
Publication Date: 
March 22, 2018

This is a guest post. The views expressed in this article are solely those of the author and do not represent positions of IEEE Spectrum or the IEEE.

Uber’s accident earlier this week—the first fatality involving a pedestrian and an autonomous car—already has fingers pointing to various parties, though it’s still too early to tell who’s responsible. The plot thickens every day with new rumors and information.

A woman, jaywalking at night. Walking a bicycle. Possibly homeless. On a busy street known for partying. In Arizona, with permissive laws. Involving Uber, with permissive ethics. Which had hired an ex-felon to be the safety driver in the autonomous vehicle (AV).

As we wait for a full investigation, we can start untangling the strands of responsibility, which include the following possibilities.

1. The victim

First, it should be clear that whether Elaine Herzberg, the 49-year old victim, was homeless or not is irrelevant, even if social prejudices against the homeless are exploited in placing blame. Hypothetically, if she had been drinking at one of the many bars nearby—next to Arizona State University, which was once Playboy’s #1 party school—that could be relevant. Jaywalking, despite nearby crosswalks, would also be relevant.

But even if the Uber car had the right of way, it can’t just run over whatever is in front of it. Imagine if it were a toddler who darted into the road: is there really no obligation to avoid the crash, just because a robot car has the right of way?  (No.) Even if it were an unavoidable crash, one can still ask whether the car should have driven more slowly, so to not put itself in such a scenario in the first place.

2.  Uber

Tempe’s police chief Sylvia Moir said, “Preliminarily, it appears that the Uber would likely not be at fault in this accident.” Even if true, this speaks only to traffic laws and not about other liability. All technologies have their limits; if Uber hadn’t properly accounted for these limits—such as not worrying that a pedestrian might be hidden behind an object, like a tree or parked truck—the company could share some responsibility. 

Even if the victim had stepped out from dark shadows, poor lighting shouldn’t be an issue given the lidar (basically, a laser-based radar) used by Uber. Though as Uber admits, some lidars are better than others. If lighting was an actual issue, then darker-skinned people might want to be extra wary, given that the best technologies today have a hard time detecting their faces.

AVs also have a hard time recognizing bicycles. If this turns out to be a factor, then it’s a known factor and should have been accounted for. These and other technology limits, especially foreseeable errors, may mean that AV technologies were not ready for public streets, and Uber oversold its capabilities to us. For many, this wouldn’t be out of character for Uber, given a perceived disregard for laws and regulations for public safety.

Read the full piece at IEEE Spectrum