Autonomous vehicles will almost certainly precipitate an unparalleled reduction in traffic fatalities. And as someone who is deeply disturbed by the sheer magnitude of carnage our society is willing to tolerate for the sake of motorists’ convenience, I absolutely welcome this change. However, while Voltaire cautions us to not allow the perfect to be the enemy of the good, we are at the precipice of an incredibly troubling convention to be embedded into the state of our transportation system. I am of course referring to Mercedes-Benz’s announcement that they will be prioritizing the lives of vehicle occupants over all other people using the street. I wish to explore the disconcerting implications of elevating the value of customers’ lives over those of bystanders, and moreover, the general decline in our collective ability to distinguish between moral and economic philosophies.
The so-called trolley problem has been a staple of introductory philosophy courses for some decades now. However this past year as autonomous vehicles have begun to pass several notable milestones, the once farcical thought experiment has developed a new dimension of near-future dystopian realism. Briefly, the trolley problem asks if it is morally correct to divert a trolley so that instead of killing multiple people on one track, it kills a single person on another track. The fundamental tension is between the number of lives saved and one’s own implication in the determination of who is killed. There are then a number of standard variants which incorporate elements such as:
- the implication of previously uninvolved parties as a means to prevent a greater death toll
- the moral character or societal value of the people who may be killed, and
- the age and health of the people who may be killed.
The plot lines of the thought experiment are often fantastical enough for one to be comfortable in failing to arrive at an suitable conclusion. And more to the point, were one to actually face such an outlandish scenario, the subject could hardly be faulted if in a split-second decision selected a suboptimal alternative. Yet, the context of autonomous vehicles forces us not only to confront our nebulous sense of morality, but to then resolve these issues and hard code those decisions into a machine which will unflinchingly execute them. In fact, MIT has begun doing just that with their own survey intended to gauge public opinion on a number of these hypothetical scenarios.
Before we even attempt to grapple with the moral philosophy that informs one’s response to the trolley problem, we must first acknowledge that such a philosophy exists independent of a purely capitalistic framework. For it is often implied that agents acting in their own self interest is itself the intrinsically moral action. And while it is true that free market exchanges may produce morally praiseworthy results, they themselves are not inherently morally praiseworthy.
Now I have no intention of accomplishing in a blog post what millennia of civilization’s brightest minds have failed to do and precisely define what constitutes a morally correct action. However, I think it should be uncontroversial to claim that there are actions that people may or may not take that will better or worsen the wellbeing of themselves and others. If this proposition can be accepted, then we can say that the study of morality is the study of these actions and how they affect the wellbeing of others. And while economic philosophy is certainly related to these outcomes, its focus is in the allocation of resources, rather than the wellbeing of people. Of course people often require these resources in order to achieve this wellbeing, which is why I claim that economic systems may produce morally praiseworthy results, but are not so themselves. One can even argue, as many undoubtedly have, that some economic systems generally lead to better or worse moral outcomes, but again: the mere act of evaluating these outcomes is distinct from the economic systems which produced them.
If I am over-emphasizing the point, it is because its subtlety often gets lost in practice. Consider the claim that an action is not personal; it’s “just business.” Here you can see that an economic justification has supplanted a moral one. One is excusing his causing a presumably negative change in another’s wellbeing by citing his own positive economic benefit. Now one might argue that he also has moral responsibilities to himself, and not just others—which is probably correct—but that is not what is being done here. For that to be the case, further inquiry would be required to determine which course of action provided a better overall outcome. In actuality, the subject has shielded further inquiry by citing an economic defense as a moral one.
But in the U.S. especially, we have a tendency to accept this slight-of-hand. It is said that polluting industries have no choice but to continue polluting in order to remain competitive. The implication of course is that by failing to remain competitive, a number of employees could be left without work, and without work these employees’ wellbeing would suffer. So while that may be true, we can see that the industry is providing both morally positive (employees’ access to resources) and morally negative (environmental and human health problems) outcomes, all hiding behind a single economic—not moral—justification.
It is my position that moral considerations ought to inform economic ones: not vice versa. But as I alluded to before, there is no clearly correct moral system that we may reference. Frankly, I am not convinced that one can exist, even theoretically. What I think can exist though is allocating the space to grapple with those moral considerations, independent of the economic ones. I do believe that in earnestly discussing these moral considerations we will tend to arrive at better outcomes than pretending that they don’t exist. Unfortunately, the collapse of these two spaces into a single one has allowed us to do just that.
Autonomous Trolley Problem
To return now to the matter at hand, Mercedes-Benz has argued that they will bypass these finicky ethical questions by simply prioritizing the wellbeing of their customers at the expense of all who are so unlucky to find themselves outside these vehicles. We can certainly touch upon some of the moral arguments embedded in this deeply complex issue, but make no mistake: the car manufacturer did not evaluate the ethical conundrum and conclude with a morally justified position. They made a business decision in order to sidestep the trouble of considering their moral consequences. I have no doubt that it is a good business decision, and many would argue that a private corporation cannot be expected to prioritize morality over profit. But this is exactly why as a society we must spend the effort required to intentionally evaluate and regulate these practices in lieu of defaulting to the financial incentives of private corporations. For as described above: the two are certainly not the same, and will quite often be at odds.
Still, despite the novelty of this particular issue, it is hard to dispute that this isn’t in fact already the state of affairs in car manufacturing. The safety of the occupants has always been prioritized over those on the outside, be they other drivers or pedestrians. And that mentality is expressed by consumers’ demand for increasingly massive vehicles, leading to an arms race of sorts. Of course the safety that is sold to the consumer comes at the expense of those who must contend with this increased presence of ever more outwardly deadly vehicles on the road. There is no reason to believe that this demand will abate as vehicles become autonomous.
Perhaps it should go without saying, but to be clear, I find it to be morally intolerable for one class of people, with access to motor vehicles, to be valued above another class of people, without. Unfortunately the economic incentives of companies align with the protection of their customers at the exclusion of all others, and regulations have failed to create a framework that would properly incentivize the protection of the vulnerable as well. Yet the fact that this is the current state of affairs does not justify its continued existence, and the transition towards autonomous vehicles presents as good an opportunity as any to reevaluate these norms.
However, establishing the equal protection of all people as the new norm does present a particularly perplexing moral dilemma. Autonomous vehicles will almost certainly be vastly safer than error-prone human drivers, so any regulation that might prevent them from reaching critical mass—say, potential car-buyers balking at the idea of their purchase choosing to sacrifice them in a cold utilitarian calculus—may lead to far worse carnage than the marginal case we are investigating. And it must be acknowledged that we are indeed discussing an edge case here. So is it worth all of the hullaballoo in order to avoid further concentrating the admittedly lower rates of traffic violence on the most vulnerable people? I would claim so, for despite the path dependency of autonomous vehicle uptake, once we arrive at that equilibrium we will likely be left to contend with whatever convention has been established.
One important distinction between the trolley problem as a philosophical exercise and a practical application is the way in which uncertainty is handled. For the sake of a thought experiment, certain outcomes to various actions provide clarity in examining the foundations of one’s beliefs. However in reality, present actions can only be understood to have probabilistic future outcomes. Whichever way the vehicle is programed, it can only control the vehicle’s actions, not the outcomes of those actions. All else equal, a vehicle occupant is far more likely to survive a collision than a pedestrian is. If a vehicle’s occupants are the sole parties whose safety it is programed to account for, as Mercedes-Benz advocates, then there is no reason for a vehicle to behave in such a way that it will sustain any damage, however minor, if avoiding such is at all possible. This includes scenarios where the vehicle will prioritize not jeopardizing its occupants in the slightest, even if it means committing actions that have high likelihood of seriously injuring or killing pedestrians. Yet the very fact of its physics means that the vehicle can sustain some degree of damage without seriously threatening the wellbeing of its occupants. So contrary to Mercedes-Benz’s stated policy, it is incumbent upon us to demand that some degree of risk be borne by vehicle occupants, even if that risk is below the threshold where serious in-vehicle injuries would be likely.
Still, this leaves us in a scenario where vehicle occupant lives are valued above others, even if for the sake of ensuring that the potential benefits of autonomous vehicles are reached. But with or without this technology, it is my belief that systemic solutions already exist that can reduce traffic violence from their current endemic level to something more akin to air and rail levels. We must therefore also consider the role of non-trolley related regulations such as limiting the speed of autonomous vehicles to less than twenty miles per hour while in urban areas, or anywhere people outside of motor vehicles are permitted to be. Such a common sense regulation would all but eliminate the possibility of collisions resulting in death and serious injury. It is difficult to imagine a scenario in which a vehicle occupant’s death would be all but certain in saving a greater number of pedestrians, when the vehicle is traveling at an appropriate urban speed.
Perhaps one could consider these suggestions as evading the underlying ethics central to the question. But while the theoretical questions are certainly interesting to contemplate, it is the practical application of them which is of most consequence to people’s lives. By failing to anticipate the regulatory framework that must exist in order to effectively integrate autonomous vehicles into the urban environment, we cede those moral discussions to the economic incentives of vehicle manufacturers. And while it would be unwise to implement regulations so restrictive that they may preclude the great safety benefits that are expected to come with the removal of fallible humans operating deadly machines, we must ensure that the remaining risk is not unduly borne by the most vulnerable people on the street. While some of this must come from requiring vehicle manufacturers’ customers to bear some risk in these narrow circumstances, the majority will come from the same strategies that we already know to reduce traffic violence. That is, reducing the speed of and interactions between any vehicles, autonomous or not. This technology certainly has great potential, but at this point it is just that: potential. How it materializes will depend entirely on how we choose to—or fail to—shape it.