The social dilemma of autonomous vehicles

Bonnefon, Jean‐François; Shariff, Azim; Rahwan, Iyad · 2016 · OpenAlex-citations

DOI: 10.1126/science.aaf2654

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper investigates the ethical programming challenges inherent in autonomous vehicles (AVs), specifically focusing on how these systems should resolve dilemmas involving unavoidable harm. The authors address a critical social dilemma: while AVs promise significant safety improvements, they must be programmed to make moral decisions in rare but inevitable crash scenarios, such as choosing between sacrificing a passenger to save multiple pedestrians or vice versa. The study aims to determine whether public moral preferences align with consumer purchasing behavior and whether government regulation of these ethical algorithms would be accepted or counterproductive. To answer these questions, the researchers conducted six online surveys between June and November 2015, recruiting US residents via Amazon Mechanical Turk. Participants were presented with hypothetical traffic scenarios requiring trade-offs between passenger safety and the greater good. The studies varied parameters such as the number of lives saved, the relationship of the passenger to the respondent (e.g., family member), and the presence of government regulation. Statistical analyses controlled for age and sex, and attention checks were used to filter invalid responses. The experimental design compared participants’ judgments on the morality of utilitarian algorithms (minimizing casualties) against their personal willingness to purchase AVs programmed with such algorithms. The findings reveal a distinct social dilemma. Participants consistently judged utilitarian AVs—those that sacrifice their passengers to save more lives—as morally superior. This preference held even when participants imagined themselves or family members in the vehicle. However, this moral approval did not translate into purchase intent. Respondents were significantly less likely to buy an AV programmed to sacrifice them for the greater good, preferring self-protective algorithms for their own use. Furthermore, participants largely disapproved of government regulations mandating utilitarian algorithms. Crucially, the study found that such regulations would drastically reduce the likelihood of purchasing an AV (median likelihood dropped from 59 to 21), potentially delaying the widespread adoption of safer autonomous technology. The significance of these results lies in the paradox they present for policymakers and manufacturers. While utilitarian programming may minimize overall casualties, enforcing it through regulation could discourage public adoption of AVs, thereby increasing total traffic deaths due to the slower transition from human-driven to autonomous vehicles. The authors conclude that there is no easy reconciliation between collective moral values and individual self-interest in this context. They argue that regulators must weigh the immediate safety benefits of AV adoption against the ethical implications of their programming, noting that public sentiment may evolve as the technology becomes more prevalent.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• acceptance adoption