The Right Way to Make Driverless Cars Safer
Rather than punishing an entire industry after a fatality, Arizona's governor took a sensible, measured approach.
On Monday, just days after a self-driving car operated by Uber struck and killed a pedestrian in Tempe, Ariz., Gov. Doug Ducey directed the Arizona Department of Transportation to suspend the ride-sharing company's testing of autonomous vehicles (AVs) in the state. Ducey's action underscores the seriousness with which developers and policymakers alike are treating the nation's first fatality resulting from an AV collision. And it brings attention to the hard questions that must be asked about what exactly went wrong with Uber's technology -- questions for which we will have answers in due time.
What's notable about Ducey's action is what he did not do: shut down all AV testing in the state. There's a lesson to be learned from Ducey's sensible, measured approach. In light of automation technology's promise, policymakers should be reluctant to lump all AV companies into a single, undifferentiated category in the wake of a failure. Indeed, as with virtually all existing transportation safety regulation, the best way forward is to continue to treat cases individually and to dole out punishments or regulatory sanctions specifically tailored to the nature, scope and severity of the given failure.
To that end, the National Traffic Safety Board (NTSB) has begun investigating what caused the crash that killed Elaine Herzberg. While it is wise to temper judgments about what exactly occurred until the results from the NTSB come back, it appears likely from initial reports that Uber's technology seriously malfunctioned. If that turns out to be the case, the firm should be held accountable.
The fact that Uber is already being held accountable in Arizona, which touts its permissive approach to AV testing and permitting, is no surprise. The state's March 1 executive order outlining the terms under which the vehicles are allowed to operate explicitly provides for meaningful regulatory consequences in the wake of a failure to adhere to safety expectations.
Still, a narrative persists that prophylactic caution, typically in the form of mountains of compliance paperwork, is the sole approach to meaningful regulatory supervision that can function as a guarantor of society's safety. This could not be further from the truth.
Taking a permissive approach to AV permitting is a viable regulatory alternative precisely because of regulators' ongoing supervision of clearly articulated safety expectations. Flexibility does not preclude rigorous enforcement, and neither element is the entire story of the structures limiting AV testing behavior. In fact, there are three other meaningful checks that define the field.
First, there is the National Highway Traffic Safety Administration, the primary federal regulator for motor vehicles. This agency possesses robust recall authority, which can take any vehicle off the road that it determines poses an unreasonable risk to consumer safety. Second, public scrutiny, which has already been brought to bear through the substantial media backlash from this accident, will continue to play a major role as Uber attempts to repair its reputation in the public eye as a safe AV operator. Third, civil liability, which functions as a mechanism to assess the extent of individualized harm, has a long tradition of shaping behavior associated with new technologies.
In light of all of those factors limiting the behavior of AV developers, those bemoaning Arizona as the "wild west" of AV testing are either making a claim relative to the incompetent flailing of regulators elsewhere or are simply unaware of the myriad strictures facing driverless-vehicle testing and deployment everywhere.
We should be careful not to overlearn the lessons of this accident and project the failures of this crash onto the operation of all AVs. By all accounts, the field is progressing rapidly. Waymo, for instance, is now so confident in its AV technology that the Google spinoff company has been operating in Arizona without a safety driver in the front seat since November with no injuries to date.
Policymakers, and the public they represent, must not forget that driverless cars offer an exciting future with fewer accidents and less-expensive mobility. But making the future happen faster means encouraging regulators to differentiate between the behavior of AV operators on a case-by-case basis. So far, Arizona's framework is doing exactly that.