Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AI Eyes on the Road: Cross-Cultural Perspectives on Traffic Surveillance

Ziming Wang, Shiwei Yang, Rebecca Currano, Morten Fjeld, David Sirkin

TL;DR

This study investigates cross-cultural public attitudes toward AI eyes on the road and the social trade-offs between safety and privacy. It employs a $3×3$ factorial online survey across China, Europe, and the USA, comparing Conventional Surveillance (CS), AI-Enhanced Surveillance (AS), and AI-Enhanced Surveillance with Public Shaming (PS), and measures four TAM-VS-based constructs: Perceived Capability, Perceived Risk, Perceived Transparency, and Acceptance. Key findings show convergence on CS across regions, while AI-enhanced modes are viewed less favorably in Europe and the USA; PS is consistently the least preferred mode, whereas Chinese respondents show higher AI acceptance and smaller mode gaps. The results highlight the central role of culture, familiarity, and contextual trust in the governance and design of AI-powered public surveillance, with implications for privacy, equity, and governance at scale.

Abstract

AI-powered road surveillance systems are increasingly proposed to monitor infractions such as speeding, phone use, and jaywalking. While these systems promise to enhance safety by discouraging dangerous behaviors, they also raise concerns about privacy, fairness, and potential misuse of personal data. Yet empirical research on how people perceive AI-enhanced monitoring of public spaces remains limited. We conducted an online survey ($N=720$) using a 3$\times$3 factorial design to examine perceptions of three road surveillance modes -- conventional, AI-enhanced, and AI-enhanced with public shaming -- across China, Europe, and the United States. We measured perceived capability, risk, transparency, and acceptance. Results show that conventional surveillance was most preferred, while public shaming was least preferred across all regions. Chinese respondents, however, expressed significantly higher acceptance of AI-enhanced modes than Europeans or Americans. Our findings highlight the need to account for context, culture, and social norms when considering AI-enhanced monitoring, as these shape trust, comfort, and overall acceptance.

AI Eyes on the Road: Cross-Cultural Perspectives on Traffic Surveillance

TL;DR

This study investigates cross-cultural public attitudes toward AI eyes on the road and the social trade-offs between safety and privacy. It employs a factorial online survey across China, Europe, and the USA, comparing Conventional Surveillance (CS), AI-Enhanced Surveillance (AS), and AI-Enhanced Surveillance with Public Shaming (PS), and measures four TAM-VS-based constructs: Perceived Capability, Perceived Risk, Perceived Transparency, and Acceptance. Key findings show convergence on CS across regions, while AI-enhanced modes are viewed less favorably in Europe and the USA; PS is consistently the least preferred mode, whereas Chinese respondents show higher AI acceptance and smaller mode gaps. The results highlight the central role of culture, familiarity, and contextual trust in the governance and design of AI-powered public surveillance, with implications for privacy, equity, and governance at scale.

Abstract

AI-powered road surveillance systems are increasingly proposed to monitor infractions such as speeding, phone use, and jaywalking. While these systems promise to enhance safety by discouraging dangerous behaviors, they also raise concerns about privacy, fairness, and potential misuse of personal data. Yet empirical research on how people perceive AI-enhanced monitoring of public spaces remains limited. We conducted an online survey () using a 33 factorial design to examine perceptions of three road surveillance modes -- conventional, AI-enhanced, and AI-enhanced with public shaming -- across China, Europe, and the United States. We measured perceived capability, risk, transparency, and acceptance. Results show that conventional surveillance was most preferred, while public shaming was least preferred across all regions. Chinese respondents, however, expressed significantly higher acceptance of AI-enhanced modes than Europeans or Americans. Our findings highlight the need to account for context, culture, and social norms when considering AI-enhanced monitoring, as these shape trust, comfort, and overall acceptance.

Paper Structure

This paper contains 36 sections, 6 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Surveillance Technology in HCI
  4. Human-AI Interaction in Daily Lives
  5. Cultural Influence on Attitudes about Privacy, Security, and Surveillance
  6. Methodology
  7. Study Design and Procedure
  8. Surveillance Modes
  9. Measures
  10. Sample and Measure Validity
  11. Results
  12. Statistical Analyses and Overview
  13. Perceived Capability
  14. Regional comparisons under mode conditions
  15. Comparisons of modes within each region
  16. ...and 21 more sections

Figures (6)

  • Figure 1: Interaction plot of Perceived Capability with error bars (95% confidence interval); simple effects are indicated with asterisks: * for p < .05, ** for p < .01, and *** for p < .001. (CS = conventional surveillance; AS = AI-enhanced surveillance; PS = AI-enhanced surveillance with public shaming.)
  • Figure 2: Interaction plot of Perceived Risk with error bars (95% confidence interval); simple effects are indicated with asterisks: * for p < .05, ** for p < .01, and *** for p < .001. (CS = conventional surveillance; AS = AI-enhanced surveillance; PS = AI-enhanced surveillance with public shaming.)
  • Figure 3: Interaction plot of Perceived Transparency with error bars (95% confidence interval); simple effects are indicated with asterisks: * for p < .05, ** for p < .01, and *** for p < .001. (CS = conventional surveillance; AS = AI-enhanced surveillance; PS = AI-enhanced surveillance with public shaming.)
  • Figure 4: Interaction plot of Acceptance with error bars (95% confidence interval); simple effects are indicated with asterisks: * for p < .05, ** for p < .01, and *** for p < .001. (CS = conventional surveillance; AS = AI-enhanced surveillance; PS = AI-enhanced surveillance with public shaming.)
  • Figure 5: The distribution of age by region.
  • ...and 1 more figures