The Revolution of Self-Driving Cars

Artificial Intelligence has changed how many actions are performed by individuals, leading to what one could call a digital revolution. Self-driving cars are part of this new revolution and appear as one of the most promising technologies in our society. The equipment of these vehicles with Artificial Intelligence systems run by sophisticated algorithms might change forever how transportation and driving are conceived by humanity. This technological advancement in the automobile industry has demonstrated the significant advantages that society will benefit from, including enhanced safety, increased mobility, and environmental impact. However, the use of self-driving cars is also set to give rise to problematic aspects which need attention and consideration before moving forward and relying entirely on this new technology. This article aims at providing a clear insight into self-driving vehicles and the main advantages and disadvantages that they might lead to. The first part of the essay will be devoted to illustrating the meaning of self-driving cars and the history behind their development. Subsequently, the essay will analyze the benefits of self-driving cars with particular emphasis added on road safety, the right to mobility, and environmental sustainability. Furthermore, the essay will illustrate the main issues currently debated which concern both ethical and legal problems. In the concluding part, final remarks will be given.

Development and Definition of Self-Driving Cars

Self-driving vehicles can be defined as autonomous systems which are capable of taking independent decisions, therefore without the need for human interference (Taeihagh and Lim, 2019). What is even more revolutionary is that they are able to make autonomous decisions in situations characterized by uncertainty and to keep learning from their decisions and mistakes (Danks and London, 2017). However, they also rely on big data and sensors to develop highly efficient systems (Long, Hanford, Janrathitikarn, Sinsley and Miller, 2007). The history of self-driving cars is not easy and as almost all great changes their development took time. The first computer-controlled vehicle was created by Carnegie Mellon University and introduced in the 1980s as Navlab 1. This self-driving vehicle was able of driving itself on the road and avoiding obstacles (Takács, Rudas, Bösl and Haidegger, 2018). Subsequently, a European-funded project led to the development of prototype cars that were capable of efficiently driving autonomously and keeping lines of traffic (Takács, Rudas, Bösl and Haidegger, 2018). The astonishing results demonstrated by the first self-driving cars urged many countries, including China, India, and the U.S. to invest in this technology leading to today’s self-driving cars (Daily, Medasani, Behringer and Trivedi, 2017).

Nowadays, autonomous vehicles are classified into five different categories according to the features they have. The Society of Automotive Engineers describes the first level of automation as assisted automation. The second level is characterized by partial automation. In both levels 1 and 2 the dynamic tasks are performed by humans (SAE, 2014). From level 3 to level 5 the role of the human driver becomes progressively less essential. At level 5 the vehicles are capable of driving autonomously and humans’ interference becomes irrelevant. The last level is, thus, the most revolutionary as it changes as driving is performed and conceived by humans. This great change leads us to the main advantages and disadvantages that should be considered when analyzing this new technology.

Figure 1: Navlab I experiment (Carnegie Mellon University).

Enhanced Safety

Recent studies demonstrate that the vast majority, up to 90%, of car accidents are caused by human errors (Smith, 2013). The shared misconception that autonomous cars are more dangerous than human drivers is perhaps due to ignorance and lack of acquaintance with these types of vehicles (Nees, 2019). Self-driving vehicles offer enhanced safety which could reduce car accidents and guarantee enhanced safety on roads given that Artificial Intelligence-based systems can easily outperform humans. In fact, they do not feel fatigue or distress, guaranteeing a steady and high level of focus while driving. Furthermore, robotic cars will offer a safe alternative for people choosing to get behind the wheel despite having consumed intoxicating substances. Furthermore, AI-based systems are able to conduct tasks more efficiently than humans, and driving would easily be conducted by automated vehicles with results expected to be far more promising than human driving (Smith, 2013). The skepticism toward robotic cars cannot, therefore, be based on safety concerns. Moreover, algorithms would be programmed to respect and oblige all road signs and speed limitations. If roads were occupied by solely autonomous cars, stop signs would always be respected and, it goes without saying, fewer accidents would be caused by incautious drivers (Stilgoe, 2021). Moreover, computers do not feel human emotions and stressful situations would not be dealt with in the same way by driverless cars. Autonomous vehicles, more precisely the algorithm on which they base their functioning, would be able to make speedy decisions without needing time to process information or have a rational reaction. Algorithms would function based on the instructions they receive and the learning they are able to conduct independently.

Increased Mobility

Another advantage linked to the use of autonomous vehicles is one which can benefit society as a whole and particularly individuals affected by motor disabilities. In a world still not completely suitable for people with disabilities and in which travel barriers represent an obstacle for them, autonomous cars could be a solution to some of their problems. More than one billion people across the globe have physical disabilities that prevent them to be independent, resulting in them not having jobs and facing economic difficulties (Dicianno et al., 2021). Studies also show that people with motor disabilities are dependent on others to travel and, in some cases, choose not to move altogether (Dicanno et al., 2021). Cities do not always offer high-quality and efficient public transportation systems and, when they do, the prices to access it might be too high (Roessler et al., 2013). New technologies and the result of new ideas should offer solutions to practical problems and thus benefit those who need them the most. Self-driving cars could help those with motor disabilities to feel and be independent in their daily lives, offering them the chance to freely when moving from one place to the other. The absence of human interference in the functioning of autonomous vehicles would entail that even those who would normally be able to drive a car, could enter one and be transported without the need to rely on others or public transportation.

Figure 2: An example of a self-driving car (Gates, 2013).

Environmental Impact

The number of people populating the earth is increasing and so will the number of drivers (United Nations, 2022). More cars on the road entail more accidents, which could be prevented with the use of self-driving cars eliminating most of the common causes of car accidents. However, one other phenomenon is endangering the lives of the human species: climate change. The world we live in is being affected by humans, their habits, and the tools they use, cars being one cause of climate change. More cars on the road lead to more traffic jams and higher carbon emissions. Recent studies have demonstrated that researchers estimate that self-driving cars could lower carbon emissions as a result of more efficient traffic flow and parking (Greenblatt and Shaheen, 2015). In fact, automated vehicles functioning on the base of Artificial Intelligent systems would not be distracted by passengers’ conversations or radio chattering and would never miss their turns, ultimately causing longer car drives. Furthermore, self-driving cars would keep to a consistent velocity without wasting gasoline. This feature, known as eco-driving, can reduce fuel consumption by up to 20% (Gonder et al., 2012). Despite the potential higher numbers of people choosing to use a car thanks to the increased mobility effects mentioned in the previous chapter, studies illustrate that the ultimate effect would be a reduction of carbon emissions (Greenblatt and Shaheen, 2015). This would lead to better air quality, with health benefits for society, and a favorable impact on the environment. Self-driving cars could also be linked to new trends such as the spread of electronic vehicles. The two technologies combined would represent an even greater change with impressive benefits for the environment (Kopelias et al., 2020).

Ethical Issues: The Trolley Dilemma

Despite the numerous advantages, the main of which is illustrated above, a number of issues are currently being debated among scholars. The ethical concerns of many scholars are also shared by numerous citizens who worry that self-driving cars might pose complex ethical questions that only a human could solve. In fact, despite the promising expectations posed by self-driving car technology, it is inevitable that accidents will occur. Accidents might be caused by careless driving behavior by a human driver and a collision with an autonomous car or even in the case of a malfunctioning of the algorithms of self-driving cars. Technology cannot be considered as a synonym for perfection and accidents are bound to happen (Nyholm and Smids, 2016). The most well-known scenario is called the “trolley dilemma”. It creates a fictitious scenario in which one must make a split-second decision on whether to sacrifice the life of one person in order to save the lives of a larger group of people (Lawlor, 2022). People often claim that a self-driving car, as an Artificial Intelligence system with no emotions or consciousness, could not make such a highly ethical decision. Therefore, it is believed that such decisions would be better made by humans who are capable of linking their thinking process to morality, principles, and emotions. However, people forget that when faced with life-or-death situations their thinking process is very likely to be irrational and driven by impulses such as survival instincts. The self-driving car will never feel stressed and will never feel overwhelmed. A rational and stable decision-making process will always be the main feature of systems based on algorithms. For this reason, it is perhaps more desirable to have a standardized solution to ethical problems which is shared by multiple stakeholders such as legislators, citizens, lawyers, engineers, manufacturers, and coders. A solution to this scenario, and others, could be provided by legislators in guidelines for producers of self-driving cars (Coca-Villa, 2018). This could end ethical concerns and help make the liability regime less confusing. However, it is also important to note that these ethical dilemmas exist in every aspect of life and that the law already provides answers and solutions to said concerns. Criminal law concepts have in fact proved to be successful in balancing different interests and perspectives. The concept of “state of necessity” already solves situations in which an individual causes harm to another one in order to preserve their safety (Arnolds and Garland, 1974). Article 54 of the Italian Criminal Code is an example of the legislative codification of said concept. Legislators already protect those who commit harm because of the context they found themselves in and provide that those people acting in necessity cannot be punished for their actions. The same concept could be used in the automotive industry and solve the so-called trolley dilemma. The only question left to answer is whether the algorithm should choose to save the life of the people inside the car or the ones outside the vehicle. The first answer seems the most logical given that people usually follow egoistical instincts. Furthermore, algorithms that choose the second option by default would dissuade people from putting their trust in self-driving cars. It is now the time for legislators to step in and provide solutions to practical problems related to ethics with appropriate guidelines.

Figure 3: The Trolley Dilemma (Awad et al., 2020).

Legal Concerns: a Confusing Liability Regime

A second concern is most spread amongst those who could be considered liable for the faults of the autonomous vehicle. As of now, Artificial Intelligent systems are not considered persons from a legal perspective and as such they do not have rights and obligations. It comes without saying that cars cannot be held liable for the events caused by them. This opens the door to a worrying scenario in which producers and sellers are disincentivized to risk economic investments and produce self-driving cars. In fact, it is very likely that people involved in the construction and distribution of the car would be held liable in the case of accidents. The responsibility would be shifted from the object to the manufacturer or the distributor as a result of a vicarious liability regime. The Product Liability Directive issued by the European Union places the responsibility for product defects on its producer (Cabral, 2020). Even though this is necessary to protect consumers, it is also concerning that a producer should be liable for the mistakes committed by a system that is capable of autonomous learning. Artificial Intelligence algorithms are in fact bound to develop their abilities further from the initial training received by the producer and the coder. Algorithms are also considered “black boxes” whose functioning is never completely understandable. Thus, legislators should perhaps think of solutions to find a better balance in protecting the interests of both producers and consumers. In fact, if innovation is not encouraged by legislators, society will not be able to develop from the current state it is now. Encouragement should then be followed by regulation to ensure that progress is aligned with ethics and principles accepted by legal frameworks. It would be desirable to impose mandatory insurance schemes that would exempt the producer to bear the costs needed to compensate the harmed individual seeking damage compensation (Cabral, 2020). However, steps forward have been made and the current proposal for an AI Regulation is currently taking its final shape and form (Schuett, 2023). The proposal provides for risk-based classification of Artificial Intelligence systems and provides greater obligations for producers of high-risk systems. As a result, consumers can expect effective protection against damages caused by Artificial Intelligence systems, including self-driving cars. However, one is left to wonder whether Artificial Intelligent Regulation solves the problem of the great liability obligations put on the shoulders of producers. The balance seems more in favor of consumers’ protection entailing a presumption of responsibility of the producer who will perhaps be less encouraged to continue developing new technologies.

Conclusions

In conclusion, this essay has illustrated the main arguments in favor and against the use and development of autonomous vehicles. New technology, based on Artificial Intelligence systems running on algorithms, would make human interference in driving unnecessary. For this reason, higher safety conditions can be expected drastically reducing the causes of car accidents. Moreover, the lack of need for a human driver would make transportation easier for a large number of people, specifically people affected by motor disabilities. The environment would also benefit from the combination of electric engines and self-driving cars. Carbon emissions would lower thanks to the efficiency of autonomous vehicles in keeping constant speed levels and efficient driving choices. However, scholars and stakeholders are still debating many of the issues regarding self-driving cars. Ethical dilemmas leave legislators and citizens worrying that self-driving cars will not be able to take ethical decisions. Furthermore, the current liability regime exposes producers and coders to bear the costs of unexpected damages caused by the always-evolving system. However, both problems could be solved with appropriate policies which balance the opposing interests at stake. Even though the road ahead is still rocky, there is nothing that a self-driving car cannot do when innovation is finally encouraged and properly regulated.

Bibliographical References

Arnolds, E. B., & Garland, N. F. (1974). The Defense of Necessity in Criminal Law: The Right to Choose the Lesser Evil. The Journal of Criminal Law and Criminology, 65(3), 289–301. https://scholarlycommons.law.northwestern.edu/cgi/viewcontent.cgi?article=5903&context=jclc

Cabral, T. S. (2020). Liability and artificial intelligence in the EU: Assessing the adequacy of the current Product Liability Directive. Maastricht Journal of European and Comparative Law, 27(5), 615–635. DOI: https://doi.org/10.1177/1023263X20948689

Coca-Vila, I. (2018). Self-driving Cars in Dilemmatic Situations: An Approach Based on the Theory of Justification in Criminal Law. Criminal Law, Philosophy 12, 59–82. https://philpapers.org/rec/COCSCI

Daily, M, Medasani, S, Behringer, R and Trivedi, M. (2017). Self-Driving Cars. Computer, 50(12), 18-23. DOI: 10.1109/MC.2017.4451204

Danks, D., & London, A. J. (2017). Regulating autonomous systems: Beyond standards. Expert Opinion, Carnegie Mellon University. http://inseaddataanalytics.github.io/INSEADAnalytics/OtherArticles/DanksLondonRegulatingAutonomy.pdf

Dicianno, B, E, Sivakanthan, S, Sundaram, A, Satpute, S, Kulich, H, Powers, E, Deepak, N, Russell, R, Cooper, R, Cooper, R, A. (2021). Systematic review: Automated vehicles and services for people with disabilities. Neuroscience Letters, 761. https://pubmed.ncbi.nlm.nih.gov/34237416/

Gonder, J, Earleywine, M, Sparks, W. (2012). Analyzing Vehicle Fuel Saving Opportunities through Intelligent Driver Feedback. SAE International. https://afdc.energy.gov/files/u/publication/53864.pdf

Greenblatt, J.B., Shaheen, S. (2015). Automated Vehicles, On-Demand Mobility, and Environmental Impacts. Curr Sustainable Renewable Energy Rep, 2, 74–81. https://link.springer.com/article/10.1007/s40518-015-0038-5

Kopelias, P, Demiridi, E, Vogiatzis, K, Skabardonis, A, Zafiropoulou, V. (2020). Connected & autonomous vehicles – Environmental impacts – A review. Science of The Total Environment, 712. https://pubmed.ncbi.nlm.nih.gov/31927439/

Lawlor, R. (2022). The Ethics of Automated Vehicles: Why Self-driving Cars Should not Swerve in Dilemma Cases. Res Publica, 28, 193–216. https://link.springer.com/article/10.1007/s11158-021-09519-y

Long, L. N., Hanford, S. D., Janrathitikarn, O., Sinsley, G. L., & Miller, J. A. (2007). A review of intelligent systems software for autonomous vehicles. EEE symposium on Computational intelligence in security and defence applications, 69-76. https://www.researchgate.net/publication/4249938_A_Review_of_Intelligent_Systems_Software_for_Autonomous_Vehicles

Nees, A, M. (2019). Safer than the average human driver (who is less safe than me)? Examining a popular safety benchmark for self-driving cars. Journal of Safety Research 69, 61-68. https://www.sciencedirect.com/science/article/pii/S0022437518304511?casa_token=ysGy0Wp1NRAAAAAA:LvAVNQfumblzZFlF31Axh905VnlbUmNBuLsNn3l5iSxxAC9JpY1uOvBfRUBczpsERSae2-Qa378#s0005

Nyholm, S., Smids, J. (2016). The Ethics of Accident-Algorithms for Self-Driving Cars: an Applied Trolley Problem?. Ethic Theory Moral Prac, 19, 1275–1289. https://link.springer.com/article/10.1007/s10677-016-9745-2

Roessler, R.T., et al., (2013). Specialized housing and transportation needs of adults with multiple sclerosis. Work, 45(2), 223–235. https://pubmed.ncbi.nlm.nih.gov/23752298/

SAE. (2021). International standard J3016 taxonomy and definitions for terms related to on-road motor vehicle automated driving systems (J3016_202104). SAE International. https://doi.org/10.4271/J3016_202104

Schuett, J. (2023). Risk Management in the Artificial Intelligence Act. European Journal of Risk Regulation, 1-19. DOI: https://doi.org/10.1017/err.2023.1

Stilgoe, J. How can we know a self-driving car is safe?. Ethics Inf Technol23, 635–647 (2021). DOI: https://doi.org/10.1007/s10676-021-09602-1

Taeihagh, A & Si Min Lim, H. (2019). Governing autonomous vehicles: emerging responses for safety, liability, privacy, cybersecurity, and industry risks. Transport Reviews, 39(1), 103-128. https://arxiv.org/pdf/1807.05720.pdf

Takács, A and Rudas, I, Bösl, D and Haidegger, T. (2018). Highly Automated Vehicles and Self-Driving Cars Industry Tutorial. IEEE Robotics & Automation Magazine, 25(4), 106-112. DOI: 10.1109/MRA.2018.2874301

United Nations. (2022). World Population Prospects. https://population.un.org/wpp/Graphs/Probabilistic/POP/TOT/900

Visual Sources