Liang Ya, Qian Lixian, Lu Yang, Bektaş Tolga
School of Economics and Management, Tongji University, Shanghai, China.
International Business School Suzhou, Xi'an Jiaotong-Liverpool University, Suzhou, Jiangsu Province, China.
Risk Anal. 2024 Dec 22. doi: 10.1111/risa.17692.
Advances in artificial intelligence (AI) are reshaping mobility through autonomous vehicles (AVs), which may introduce risks such as technical malfunctions, cybersecurity threats, and ethical dilemmas in decision-making. Despite these complexities, the influence of consumers' risk preferences on AV acceptance remains poorly understood. This study explores how individuals' risk preferences affect their choices among private AVs (PAVs), shared AVs (SAVs), and private conventional vehicles (PCVs). Employing a lottery experiment and a self-reported survey, we first derive four parameters to capture individuals' risk preferences. Based on a stated preference experiment and the error component logit model, we analyze reference-dependent preferences for key attributes of PAVs and SAVs, using PCVs as the reference. Our analysis reveals that risk-tolerant consumers are more inclined toward PAVs or SAVs. Further, consumers exhibit a greater sensitivity to losses, such as higher purchasing prices and running costs, than to gains, such as reduced egress time. Specifically, for buying a PAV, consumers are willing to pay 3582 CNY more for 1000 CNY saving on annual running cost, 3470 CNY for a 1-min reduction in egress time, 28,880 CNY for removing driver liability for crashes, and 30,710 CNY for the improved privacy data security. For adopting SAVs, consumers are willing to pay 0.096 CNY extra per kilometer for a 1-min reduction in access time and 0.033 CNY extra per kilometer for a 1% increase in SAV availability. Therefore, this study enhances the understanding on risk preferences in AV acceptance and offers important implications for stakeholders in the AI-empowered mobility context.
人工智能（AI）的进步正在通过自动驾驶汽车（AV）重塑出行方式，而自动驾驶汽车可能会带来技术故障、网络安全威胁以及决策中的伦理困境等风险。尽管存在这些复杂性，但消费者的风险偏好对自动驾驶汽车接受度的影响仍知之甚少。本研究探讨了个人的风险偏好如何影响他们在私人自动驾驶汽车（PAV）、共享自动驾驶汽车（SAV）和私人传统汽车（PCV）之间的选择。通过彩票实验和自我报告调查，我们首先得出四个参数来衡量个人的风险偏好。基于陈述偏好实验和误差成分logit模型，我们以传统汽车为参照，分析了对私人自动驾驶汽车和共享自动驾驶汽车关键属性的参考依赖偏好。我们的分析表明，风险容忍型消费者更倾向于选择私人自动驾驶汽车或共享自动驾驶汽车。此外，消费者对损失（如较高的购买价格和运营成本）的敏感度高于对收益（如减少出行时间）的敏感度。具体而言，对于购买一辆私人自动驾驶汽车，消费者愿意为每年运营成本节省1000元多支付3582元，为出行时间减少1分钟多支付3470元，为免除碰撞事故中的驾驶员责任多支付28880元，为改善隐私数据安全多支付30710元。对于采用共享自动驾驶汽车，消费者愿意为接入时间减少1分钟每公里多支付0.096元，为共享自动驾驶汽车可用性提高1%每公里多支付0.033元。因此，本研究增进了对自动驾驶汽车接受度中风险偏好的理解，并为人工智能驱动的出行背景下的利益相关者提供了重要启示。
