Perceptions dominate our actions either for personal, corporate, or governmental decisions. This is particularly true when decisions involve complex technologies as suggested by Chauncy Starr, Paul Slovic and colleagues. The adoption and potential regulation of autonomous vehicles faces several challenges: it is a highly disruptive technology affecting the core of our social fabric; it is a highly networked system lowering the probability but increasing the failure consequences; it will lead to changes in the structure of work and the form of cities; it ultimately require massive public investments in infrastructure to support the system; the time frame for sequential decisions is long, on the order of half a century. In many respects the challenge presented by autonomous vehicles is as daunting as that posed by global warming. At the same time, the need is great in the US and worldwide, particularly among vulnerable pedestrians and cyclists, with over a million fatalities per year and many more injuries.
In this paper we explore the way the objectives of the different sets of stakeholders (consumers, manufacturers, local and federal governments) and identify the assumptions that each group makes. Next, we make the case for market failure. Individuals will often purchase technologies that may save their own lives, but are less likely to purchase it when it involves saving someone else. We then identify risks that are faced by individual stakeholders and how cognitive biases and their own unique approaches for coping with risk affect the strategies that they adopt. Key to our approach is the separation of safety benefits of autonomous vehicle technology from other benefits such as improvements in fuel efficiency, congestion and freeing the driver from maneuvering tasks. We end by concluding that shorter term goals are more likely to be achievable and the adoption of modest vehicle autonomy are more likely to have better long term safety performance than waiting for autonomous vehicle technology to be perfected.