The Autonomous Vehicle and the Trolley Problem
An engineering firm is programming a self-driving car's emergency policy. In a rare failure mode, the car can either stay in its lane and strike two pedestrians who have entered the road illegally, or swerve and strike one pedestrian standing on the pavement. The car's passenger is unharmed in both scenarios. The firm's lawyers note that swerving toward the pavement may create legal liability, while staying in lane follows the road rules. A philosopher on the panel, Dr. Mensah, argues that the car should minimize harm regardless of legality. Another member, Dr. Ito, argues that the public will not accept a car programmed to swerve into pedestrians and that trust matters. A third asks whether the passenger's consent to the risk matters at all.
1. Compare a consequentialist and a deontological analysis of the swerve decision, and state what each implies for the programming policy.
Read the worked answer
A consequentialist analysis counts lives: swerving saves one life at the cost of one, so it minimizes harm and should be programmed. A deontological analysis asks whether the act treats anyone merely as a means or violates a duty. Staying in lane follows the rules and does not aim at the pedestrians; swerving deliberately redirects harm onto a person who was not posing a threat, which some deontologists treat as impermissible even when it reduces total harm.
The programming implication differs. Consequentialism favors a harm-minimizing rule; deontology favors a rule that avoids deliberate redirection, possibly with exceptions. Neither analysis is complete without an account of uncertainty: the car's sensors may be wrong, so the policy must be robust across error, which may favor rule-following in practice even for a consequentialist.
2. Dr. Ito raises public acceptance. Explain how trust considerations bear on the moral question, and whether they can justify a policy that does not minimize harm.
Read the worked answer
Trust affects consequences: if the public believes cars will sacrifice passengers or swerve at pedestrians, adoption may fall, and more road deaths may result overall. So trust is relevant to a consequentialist calculation, not merely a political constraint. It can also matter deontologically, since a policy that deceives the public about how the car behaves treats them with disrespect.
Trust can justify a non-harm-minimizing policy only if the expected costs of lost trust outweigh the expected harm saved. That is an empirical question, not a moral axiom. Dr. Ito's argument is therefore conditional: it succeeds if the trust effects are large and predictable, and it fails if the public can be informed and reassured. The panel should ask for evidence rather than assume either way.
3. The third member asks whether the passenger's consent to risk matters. Assess its moral relevance.
Read the worked answer
Consent matters because the passenger chooses to travel knowing the car's general policy, while the pedestrians do not choose to be exposed. This asymmetry supports a duty of care toward those outside the vehicle. But consent is thin: passengers rarely read the policy, and the pedestrians cannot consent at all, so consent cannot do all the work.
A defensible position treats consent as one factor among several. It may justify a policy that protects outsiders more than passengers, since passengers accept some risk by entering the car. It does not justify ignoring the pedestrians' interests, and it does not settle the swerve question, which turns on the permissibility of redirection. The panel should state the policy publicly so that consent is informed rather than fictional.