Write a philosophy essay comparing how utilitarianism and Kantian deontology would handle the crash decisions of a self-driving car. Keep it between 600 and 800 words. Reply with only the finished text, with no preface or closing remarks.
Text
Programming the Trolley: Utilitarian and Kantian Responses to Self-Driving Crash Dilemmas
Self-driving cars force ethical theory into code. An unavoidable crash presents a modern trolley problem: the vehicle must choose among harms. Utilitarianism and Kantian deontology offer sharply different decision procedures.
Utilitarianism begins with consequences. The morally right action is the one that produces the greatest net happiness or welfare, impartially considered. In a crash algorithm, this implies probabilistic cost-benefit analysis. If swerving left kills one pedestrian but continuing kills five, the car should swerve, because five deaths are worse than one. If turning into a wall kills the passenger but saves several schoolchildren, a strict act-utilitarian may program that choice. The passenger’s interests count equally with pedestrians’; no one’s status as owner or occupant gives extra weight. The algorithm would estimate injury severity, number of victims, and other factors only insofar as they correlate with future welfare. It would aim to minimize expected casualties and suffering. This has appeal: it is transparent, mathematically tractable, and refuses moral favoritism. But it has troubling implications. It can justify sacrificing the vulnerable if their deaths produce less aggregate welfare, and it can turn the car into an instrument of social engineering. It also demands comparisons that may be impossible: is one death worse than five broken legs? Whose welfare counts, and for how long? And if every car is programmed to sacrifice its passenger for the greater good, people may refuse to buy them, undermining the safety benefits. A rule-utilitarian might therefore endorse constraints like “never actively target pedestrians” because such rules maximize long-term trust and compliance. Yet the core logic remains aggregative.
Kantian deontology rejects aggregation. Persons are not containers of utility but ends in themselves, possessing dignity rather than price. The categorical imperative requires that maxims be universalizable and that humanity never be treated merely as a means. A crash algorithm, then, cannot deliberately kill an innocent person to save others, even if the numbers favor it. To swerve into one pedestrian because doing so saves five is to use that pedestrian as a tool for others’ survival. It treats a rational agent as a mere means, violating the Formula of Humanity. The same applies to sacrificing the passenger. The passenger did not consent to be killed for pedestrians; the car would be using him or her as a means to a social optimum. Kantian morality would instead insist on constraints: do not intentionally harm innocent people, do not discriminate among lives as if some are worth more, and do not turn the vehicle into a weapon. This does not mean the car may do nothing. It may brake, warn, deploy safety systems, and take evasive action that does not intentionally target anyone. If a collision is unavoidable, a Kantian might prefer a rule that refuses to choose among victims, accepting that tragedy may be unavoidable but refusing to make one person’s death a chosen instrument. Some Kantians might allow unintended side effects under the doctrine of double effect, but the intention must remain avoidance of harm, not the killing of a selected victim.
The contrast is stark. Utilitarianism asks, “Which outcome is best overall?” and permits harming some if the total is improved. Kantian deontology asks, “What may I do to a person?” and forbids using anyone merely as a means. In self-driving cars, the utilitarian approach promises fewer total injuries but risks justified sacrifice and public distrust. The Kantian approach protects individual dignity and equality but can appear absolutist when every option kills. A practical design might blend them: use utilitarian reasoning for risk reduction, braking distance, and route safety, while adopting deontological constraints against deliberate targeting. But theory matters because it shapes what the car is permitted to intend. The self-driving car is not merely a machine; it is a moral agent by proxy, and its code will encode an answer to the oldest question: whether ethics is about maximizing good outcomes or respecting persons, whatever the consequences.