How safe is safe enough? This is a question which has perplexed the autonomous vehicle (AV) industry since its inception. Every time I read about it, I am always left reflecting on what a dumb question this is, because the answer is easy — though I don’t think you are going to like it. Join me as we dive into that thought a little deeper.
The answer to the question — to the great question — is 42. Douglas Adams fans everywhere and lovers of The Hitchhiker’s Guide to the Galaxy will know that I am referring to the wisdom of a computer called “Deep Thought.” If you’ve no idea what I am talking about, this trip back in time will bring you up to speed.
How safe is safe enough for AVs operating on public roadways? In the words of Deep Thought: “You have to know what the question actually is in order to know what the answer means.” Safe enough for whom? Passengers? Vulnerable road users? Safe enough where? When? In what conditions? Deep Thought has the answer to all these questions: “Tricky.”
The metric I hear quoted most often is that AVs must be safer than human drivers. I’ve heard twice as safe and I’ve heard ten times as safe. We don’t know what “safe enough” means for AVs, but we know it must definitely be a multiple of whatever can be achieved by human drivers.
The statistic I see quoted most often is that there is on average approximately one fatality for every one hundred million miles of human driving. Many AV proponents say that that is the metric to measure against. I say not so fast.
Early on in my analyst career I was cautioned of the risks of statistical averages with the warning that the average person has one breast and one testicle. While statistically valid, that statement describes precisely nobody. Which reminds me of the saying that there are only two types of statistic: Those you look up and those you make up.
In researching this article, I came across a research paper called “How Safe is Safe Enough? A Psychometric Study of Attitudes Towards Technological Risks and Benefits” in which the authors quote Chauncey Starr, an American electrical engineer who was an expert in nuclear energy.
“…the ‘revealed preference’ method advocated by Starr* is based on the assumption that, by trial and error, society has arrived at an ‘essentially optimum’ balance between the risks and benefits associated with any activity.”
Thus, how safe is safe enough isn’t a question to be answered on our behalf by a bunch of self-interested tech companies salivating over their slice of a future $7 trillion “passenger economy.”
Were we to borrow a festive metaphor and “follow the Starr,” I would propose that a more informative question might be: How could society arrive at an essentially optimum balance between the risks and benefits associated with AVs operating on public roadways? We don’t yet have a definitive answer, but recent developments just took an important step in that direction.
Stop! Advocates time
If you hadn’t heard of Advocates for Auto and Highway Safety (Advocates), you have now. At the end of November, Advocates published the “AV Tenets” to guide federal legislation and policy on the development and deployment of AVs.
The press release states:
“The Tenets are divided into four main categories: prioritizing the safety of all road users; guaranteeing accessibility for all; preserving consumer and worker rights; and, ensuring local control and sustainable transportation.
“The AV Tenets are supported by a growing coalition of more than 55 groups. They are based on expert analysis, real-world experience and public opinion. A survey commissioned by Advocates in January 2020 found that nearly 70 percent of respondents said they would feel more comfortable about AVs if they knew manufacturers had to meet minimum performance standards before selling them to the public.”
This video presents the briefing in full, starting with Cathy Chase, President for Advocates, outlining the AV Tenets in the first six minutes. I invite you to take the time to watch it.
The Tenets are supported by a wide range of stakeholders covering a diverse cross-section of society. Stephen W. Hargarten, MD, associate dean, Office of Global Health, Medical College of Wisconsin, commented:
“Many have touted autonomous vehicles as a potential solution. I share this hope; however, if AVs are left unfettered, unchecked, and unregulated, this so-called ‘cure’ could result in even more preventable crashes, injuries and fatalities. This threat is especially concerning when it comes to vulnerable road users like pedestrians and bicyclists, who lack the protections afforded by a 4,000-pound car.”
As reported by Slate, that unfettered, unchecked, and unregulated policy vacuum has enabled some automakers to introduce ADAS packages with confusing names and inadequate driver monitoring. “Tesla seems to repeatedly sacrifice safety in order to gain a competitive edge,” observed David Zipper.
Recent industry developments show us that the testing, validation and deployment of AVs isn’t solely about technology. The work of Chauncey Starr highlights that the debate must also encompass ethical, moral and political viewpoints, precisely because it is society as a whole that will arrive at the essentially optimum balance between the risks and benefits associated with operating AVs on public roadways. Thanks to the AV Tenets, this “societal conversation” has at last begun.
As proposed by Advocates, and corroborated by EE Times, the short-term answer to reducing traffic fatalities in the U.S. is legislation that will advance the adoption of proven safety technology to make human drivers into safer drivers. This includes automatic emergency braking, lane departure warning and blind spot detection, coupled with vision-based driver monitoring to detect for distraction, drowsiness and impairment.
It turns out that the answer to the great question of “How safe is safe enough?” has more to do with common sense and open, informed, debate than Deep Thought. Thank you, Douglas Adams, and perhaps one day a popular automaker might release the Ford Prefect in his memory.
*Starr, C., “Social Benefit versus Technological Risk”, Science Vol. 165, pp. 1232-1238