Being good (at driving): Characterizing behavioral expectations on automated and human driven vehicles

2025-02-12

AI Summary

This study addresses the challenge of defining and evaluating the reasonableness of autonomous vehicle (AV) and human driving behaviors by proposing the “Drivership” theoretical framework. Grounded in reciprocal expectations among road users, Drivership establishes a three-layer dynamic expectation model—empirical, normative, and facilitative—and introduces the novel, value-sensitive concept of “right-of-way citizenship,” distinguishing safety-centered from socially aware driving behaviors. Methodologically, it integrates normative expectation theory, Value-Sensitive Design (VSD), behavioral modeling, and systems-level evaluation. The contributions are threefold: (1) providing an actionable theoretical foundation for AV safety assurance; (2) enabling multi-dimensional benchmarking of driving performance; and (3) shifting AV evaluation paradigms from technical compliance toward societal acceptability—thereby fostering synergistic co-evolution of ethical integration and transportation governance. (149 words)

Technology Category

• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness
• Humans and AI: Learning Human Values and Preferences
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Economics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems

Abstract

For over a century, researchers have wrestled with how to define good driving behavior, and the debate has surfaced anew for automated vehicles (AVs). We put forth the concept of Drivership as a framing for the realization of good driving behaviors. Drivership grounds the evaluation of driving behaviors in the alignment between the mutualistic expectations that exist amongst road users. Leveraging existing literature, we distinguish Empirical Expectations (i.e., reflecting"beliefs that a certain behavior will be followed,"drawing on past experiences) (Bicchieri, 2006); and Normative Expectations (i.e., reflecting"beliefs that a certain behavior ought to be followed,"based on societally agreed-upon principles) (Bicchieri, 2006). Because societal expectations naturally shift over time, we introduce a third type of expectation, Furtherance Expectations, denoting behavior which could be exhibited to enable continuous improvement of the transportation ecosystem. We position Drivership within the space of societal Normative Expectations, which may overlap with some Empirical and Furtherance Expectations, constrained by what is technologically and physically feasible. Additionally, we establish a novel vocabulary to rigorously tackle conversations on stakeholders' expectations, a key feature of value-sensitive design. We also detail how Drivership comprises safety-centric behaviors and what we term socially-aware behaviors (where there are no clear safety stakes). Drivership supports multiple purposes, including advancing the understanding and evaluation of driving performance through benchmarking based on many criteria. As such, we argue that an appropriate framing of the notion of Drivership also underpins the overall development of a safety case. The paper explores these applications under the more general tenet of Drivership as a central element to roadway citizenship.

Problem

• Define good driving behavior for AVs
• Introduce Drivership for societal expectations
• Benchmark driving performance via Drivership

Innovation

• Defining Drivership for AVs
• Introducing Furtherance Expectations
• Safety-centric and socially-aware behaviors