PUBLIC OPINION REGARDING AUTONOMOUS VEHICLE PRIVACY AND SECURITY

Abstract

 

The purpose of this thesis was to investigate privacy concerns related to autonomous vehicles with a focus on people’s level of comfort with autonomous vehicles collecting different types of information and sharing that information with different entities.  A survey of 154 United States residents was conducted through Amazon’s Mechanical Turk crowdsourcing survey tool to probe opinion regarding these issues and model trends within the sample. The findings of this study indicate that there is concern over information privacy in autonomous vehicles and that people would be hesitant about purchasing a vehicle that used their personal information in certain ways.  Study participants showed the most comfort sharing non-personally identifying information and sharing locational information, but the least comfort with sharing personally identifying information.  People were the least willing to share information with marketers and most willing to share information in emergency situations.  Respondents also indicated that they would like control over their data in terms of being able to remotely erase sensitive information and they are concerned about security of their information.  To attempt to alleviate concerns and address privacy issues, a series of recommendations for education efforts and data collection techniques were proposed for automakers and policymakers.

TABLE OF CONTENTS

LIST OF FIGURES ……………………………………………………………………………………………. vii

LIST OF TABLES …………………………………………………………………………………………….. viii

Chapter 1  Introduction …………………………………………………………………………………………. 1

Chapter 2 Literature Review ………………………………………………………………………………….. 5

Overview of Privacy and Security Issues …………………………………………………………….. 5

Consumer Privacy Preferences …………………………………………………………………………… 7

What types of information are people willing to share? ……………………………………… 8

Control of Personal Information ……………………………………………………………………. 11

Consumer Preferences of Who Can Access Personal Information ……………………… 13

The Effects of Privacy Concerns …………………………………………………………………… 15

Summary ……………………………………………………………………………………………………. 16

Chapter 3 Methodology ………………………………………………………………………………………. 18

Use of Online Surveys …………………………………………………………………………………….. 18

Mechanical Turk …………………………………………………………………………………………….. 20

Survey …………………………………………………………………………………………………………… 21

Data Analysis …………………………………………………………………………………………………. 24

Data Cleaning……………………………………………………………………………………………… 24

Analysis Methods………………………………………………………………………………………… 25Chapter 4 Survey Results …………………………………………………………………………………….. 32Respondent Information…………………………………………………………………………………… 32

Level of Concern about Personal and Information Privacy when Purchasing an

Autonomous Vehicle ………………………………………………………………………………………. 37

Preferences about Use of Credit Card Information ………………………………………………. 39

Preferences about Personally and Non-Personally Identifying Information ……………. 40

Preferences about Locational Privacy ………………………………………………………………… 43

Preferences about Information Storage and Security ……………………………………………. 45

Preferences on Remote Control of Vehicles ……………………………………………………….. 49

Chapter 5 Discussion and Analysis……………………………………………………………………….. 52

Privacy as a Concern ……………………………………………………………………………………….. 52

Locational Privacy ………………………………………………………………………………………….. 53

Type of Personal Information …………………………………………………………………………… 54

Use of Personal Information …………………………………………………………………………….. 56

Responses by Demographic Group ……………………………………………………………………. 57

Gender ……………………………………………………………………………………………………….. 63

Age ……………………………………………………………………………………………………………. 63

Income……………………………………………………………………………………………………….. 64

Education …………………………………………………………………………………………………… 65

Effects of Sample Size …………………………………………………………………………………. 65Chapter 6 Implications and Suggestions ………………………………………………………………… 67Chapter 7 Conclusions ………………………………………………………………………………………… 70

References …………………………………………………………………………………………………………. 73

Appendix A Autonomous Vehicle Survey …………………………………………………………….. 77

Appendix B Logistic Regression Models for Personally and Non-Personally Identifying

Information ……………………………………………………………………………………………………….. 92

Appendix C Logistic Regression Models for Locational Privacy ……………………………… 94

Chapter 1  Introduction

Research and discussion about autonomous vehicles has been on the rise in recent years following on-street vehicle testing in cities across the United States.  In 2015, Tesla became the first automaker to release an autonomous driving feature to the public on its Model S sedan.  Since then, there have been countless advances to these systems, and many other automakers have joined the race to release their own autonomous vehicles.  Today there are over 250 companies involved in the race towards manufacturing and selling autonomous cars, and many of the top competitors have made announcements about their plans to release fully autonomous vehicles in 2020 and 2021 (Stewart, 2017; Walker, 2017).

With companies racing to be first in putting out fully autonomous vehicles, there has been little time to conduct research on many of the issues affecting them.  Companies are primarily focused on the programming and systematic needs necessary to getting vehicles on the road, such as hard-coded rules of the road, obstacle avoidance, and object discrimination, while many other critical questions remain unanswered (“Self-Driving

Cars Explained,” 2017).  Among these important topics is data privacy and security.  While many news articles and blog posts have discussed potential uses of personal information in autonomous vehicles, little is known about whether automakers are considering the ethical implications of personal privacy and how they are approaching data collection, use, storage, and security in the designs of their vehicles.

A variety of privacy issues derived from the collection of locational data and personal information have been discussed in the media.  Vehicles are expected to collect a significant amount of locational data, primarily logging all of the destinations where the vehicle stops along with linked information, such as time of trips and order of connected trips.  In addition, it is expected that vehicles will have access to a variety of personal information ranging from non-personally identifying information (non-PII) such as driver age, gender, and race to personally identifying information (PII) such as name, home address, and credit card number (Navetta, Segalis, & Kleiner, 2017).  Some have even suggested that autonomous vehicles will comb through personal emails and listen to conversations between passengers or those “overheard” from cellphones (LaFrance, 2016).  Combinations of all of these pieces of information are then expected to be used for many different purposes.  Locational data alone can be used to determine information about drivers, such as their preferred stores or restaurants, places of work, schools, or doctors.  These data can also be used to determine commuting habits, and mass collection of this information can be used for traffic studies, planning, and routing.

Privacy concerns are not necessarily about the collection of all of these data mentioned above, but about whom this information will be given to, what will be done with it, and how it will be protected.  The data listed above will be of interest to countless groups of people.  Marketers could use the data to target specific drivers with advertisements based on their habits, preferences, and locations.  The government could use these data for surveillance purposes, law enforcement, targeted legislation, or tolling.  Engineers could use these data for congestion management; automakers could use the information for new vehicle developments; insurance agencies could use these data for policy formation; and friends and family could use these records for tracking.  While all of these possibilities have been discussed in the popular press, it is likely that there are countless other uses that have not yet been thought of, and it is unlikely that all possible scenarios will be thought of prior to the wide-spread release of autonomous vehicles.

Unless legislation is enacted, it will likely be up to automakers to decide what to do with the information collected in their vehicles.  While the media has hypothesized some good and some worrisome uses of personal information, no formal research has been done into public perception of these privacy issues as they relate to autonomous vehicles or public preferences when it comes to sharing certain types of information with certain groups.  Past research on privacy has shown that consumers have varying preferences about sharing information depending on the type of information and with whom they are sharing.  People are generally more concerned about sharing PII than non-PII and have shown varying degrees of trust in different entities with more trust in groups they conduct business with (e.g., telephone companies, credit card companies, and cable companies) and less trust in entities that simply collect information about users (e.g., search engines, online video sites, and social media sites) (Madden & Rainie, 2015).  Research has also shown that people feel they have little control over how their personal information is collected and used.

The objective of this study is to develop baseline measures of public opinion on privacy and security issues as they relate to autonomous vehicles.  The findings of this study will expand upon existing literature by applying privacy preferences directly to the potential autonomous vehicle uses mentioned above.  An up-to-date knowledge of peoples’ privacy preferences surrounding autonomous vehicles will be very important for automakers in designing autonomous vehicle systems.  Automakers will need to consider what types of information to allow their vehicles to collect, how to store the information, and whom to provide the information to.  If public opinion is not accounted for, people may not readily accept autonomous vehicles into society and may decide not to purchase vehicles for personal use.

The survey conducted for this thesis sought answers to the following questions:

• How concerned are people about information privacy and security with autonomous vehicles?
• What types of information are people most and least comfortable sharing?
• For what purposes are people most comfortable sharing information?
• With whom are people most comfortable sharing information?
• How do people want their information secured?
• How much control do people want over their information?

Using the results of the survey, this thesis develops models for predicting public opinion on autonomous vehicle privacy issues based upon demographic information and formulates recommendations for incorporating public opinion into autonomous vehicle design.  Additional recommendations are made about potential policy implications and education initiatives to protect peoples’ privacy.

PUBLIC OPINION REGARDING AUTONOMOUS VEHICLE PRIVACY AND SECURITY

Sharing is caring!