Futuramas of the present: the "driver problem" in the autonomous vehicle sociotechnical imaginary

The paper introduces the concept of a sociotechnical imaginary, using autonomous vehicles as a prime example. It highlights the prevalent narrative that autonomous vehicles will drastically improve road safety by eliminating human error, a claim often supported by statistics suggesting human error is responsible for over 90% of accidents. The authors challenge this narrative, arguing that the focus should shift from predicting a utopian future to analyzing the present-day discourses and implications of the autonomous vehicle imaginary. They argue that the autonomous vehicle is not a transformative technology but rather the latest iteration of a series of automobility imaginaries that have consistently promised utopian futures that have never materialized. The primary claim of the paper is that the statistic of 90% of accidents being attributed to human error is a flawed construct rooted in the methodologies of road safety research and has been accepted as a fact, thereby shaping the sociotechnical imaginary. The paper asserts that road violence is not merely a contingent result of human error but is intrinsic to the system of automobility itself.

The authors review existing critical social scientific research questioning the optimistic claims surrounding autonomous vehicles. This research emphasizes that autonomous vehicles are primarily social artifacts, not just technological ones, and that roads are social spaces not adequately understood through an engineering lens. The authors highlight the work of Jasanoff and Kim (2009; Jasanoff, 2015) on sociotechnical imaginaries and how they shape visions of the future. They also refer to other studies critical of the assumptions embedded in the autonomous mobility sociotechnical imaginary, pointing out the tensions between governance and engineering narratives versus citizen-focused ones. The paper acknowledges the valid concerns raised by this previous research but argues that focusing solely on the future accepts the premise that autonomous vehicles will create a radically different future. The authors advocate for a focus on the present-day significance of the autonomous vehicle sociotechnical imaginary.

The paper employs a critical analysis of the construction of road accident statistics. The authors trace the origins of the widely cited statistic that 93% of accidents are due to human error, primarily focusing on the Tri-Level Study (Treat et al., 1977a, 1977b, 1979) and subsequent studies by the US National Highway Traffic Safety Administration (NHTSA) and European organizations. They examine the methodologies employed in these studies, highlighting the underlying assumptions and theoretical frameworks. The analysis focuses on how these studies define "accidents," "human error," and "causality." The authors highlight the lack of consideration given to automobility itself as a causal factor, as opposed to drivers, vehicles, and the environment. They delve into the process of data collection, coding, and analysis, arguing that the very definition and categorization of events contribute to the outcome. The authors discuss accident causation theory, tracing its development from early linear models to more complex systems approaches, noting how these models consistently focus on human error as the primary causal factor, neglecting the systemic aspects of automobility. They contrast this approach with a view of accidents as revelations of the inherent violence within the system of automobility. This critical analysis utilizes concepts from sociology, science and technology studies, and philosophy to understand how the statistic has been constructed and its implications.

The central finding is that the statistic of 93% of road accidents being attributed to human error is not a neutral, objective measurement but a construct embedded within specific research methodologies and underlying assumptions. The authors demonstrate how these methodologies inherently exclude automobility itself as a causal factor, focusing instead on the driver, vehicle, and environment. This process of constructing the statistic is shown to involve choices in defining variables, developing codebooks, collecting data, and interpreting results. The authors highlight the consistency in methodology across various studies, leading to the repetitive confirmation of the 93% figure. They also analyze how the statistic is used rhetorically to promote autonomous vehicles as a solution to road safety problems. The authors argue that this focus on "human error" as the primary cause of accidents serves to maintain the status quo of automobility, deflecting criticism of the system as a whole and preventing a more fundamental questioning of automobility's role in road violence. They draw an analogy to firearm fatalities, arguing that while people kill people, the ease and efficiency with which killing is possible is directly related to the presence of the firearm itself. Similarly, the design and nature of automobiles, and the system of automobility, make road accidents more probable, regardless of driver behavior. The analysis shows how the conceptualization of accidents within a DVE (Driver-Vehicle-Environment) framework systematically locates causality within individual components and ignores the systemic nature of automobility and its implications for road violence. The focus on the driver and the inherent problems of human fallibility reinforces the view of technological solutions, such as self-driving cars, as a necessary and inevitable path forward. This, in turn, reinforces the existing automotive regime rather than challenging the systemic issues underlying road fatalities.

The findings challenge the underlying assumptions of the autonomous vehicle sociotechnical imaginary, demonstrating how a seemingly objective statistic serves to reinforce the existing automobility regime. The authors show that the focus on technological solutions, such as self-driving cars, distracts from the need for broader societal changes that could address the systemic causes of road violence. The emphasis on driver error as the primary cause of accidents allows the automobile industry, and the states that support it, to continue to profit from the system. The paper's discussion points to the need for alternative approaches, such as phenomenological studies that focus on the situated practices and experiences that contribute to road accidents, rather than purely focusing on a causal chain analysis. This alternative approach would broaden the scope of inquiry and consider automobility itself as a crucial element of road violence.

The paper concludes that the autonomous vehicle sociotechnical imaginary, while promising a technologically advanced future, ultimately serves to reproduce and expand the existing automobility regime. The authors highlight the artificial construction of the statistic around human error in accidents as a central element of this perpetuation. They advocate for a shift in focus towards alternative sociotechnical imaginaries that are not rooted in technological determinism but engage with the broader social and political implications of automobility and road violence. Future research should explore alternative methodologies and theoretical frameworks for understanding road accidents, considering the systemic, rather than merely individual, aspects of the problem.

The authors acknowledge that their analysis focuses primarily on the construction of road accident statistics and the sociotechnical imaginary surrounding autonomous vehicles. While they analyze the methodology of accident causation studies in detail, they do not conduct original empirical research on accident causation. The paper's focus on the Western context may limit the generalizability of some findings to other cultural contexts with different road safety norms and infrastructures. The critical approach taken in the paper may not fully address the potential benefits of autonomous vehicles in specific circumstances or the complexities of technological innovation.