User evaluations of moral behavior in self-driving cars

Abstract

In this research project we will investigate how the public thinks a self-driving car should behave when dealing with moral dilemmas that could occur while driving. We will determine to what extent people think possible algorithms for such a car behave in a fair way and whether or not these scores are a good predictor for the willingness to pay for the use of such a car. We will also try to determine the underlying factors for their scores on fairness so they can be used to predict these scores. This is done through the presentation of several scenarios containing some kind of moral conflict based on a number of different factors, followed by survey questions. Using this data, we will provide recommendations for the computational implementation of moral choices in self-driving cars. Finally, we will discuss potential ethical issues surrounding these implementations.