Level 4等級自駕車道德困境決策行為初探：電車困境的應用

A Preliminary Study on the Moral Dilemma Decision Behavior of Level 4 Self-driving: An Applied Trolley Problem

道路運輸已由輔助駕駛進入Level 4等級自動駕駛的階段，未來，駕駛人與自駕車的關係將會是「人機合作」，人類與機器「可以」同時也「必須」緊密的共同合作來完成駕駛工作。本文以Level 4等級自駕車為研究場域，探討駕駛員在啟動自動駕駛後遇到緊急情況時的決策行為。研究結果發現，當啟動Level 4自動駕駛並發生追撞情境時，多數受訪者會選擇將行車控制權交給人工智慧；而選擇由人工智慧執行針對該情境所設定的駕駛決策時，該決策多數會選擇追撞有戴安全帽的騎士，這種決策雖可降低傷亡，但卻導致有戴安全帽的騎士受到傷害。這種達到降低人類傷亡的目標卻可能變相處罰遵守交通規則的道德矛盾，讓我們意識到有必要重新審視交通法規與人類安全之間的內涵。

In recent years, road transportation has entered conditional driving automation (Level 4) from partial driving automation (Level 2). In the future, the relationship between drivers and Level 4 self-driving cars will be ''human-machine cooperation''. Humans and machines ''can'' and ''must'' be close to work together to complete the driving task. This article uses Level 4 self-driving cars as the research situation to explore the driver's decision-making behavior when encountering an emergency after starting automatic driving. The results found that when autopilot is activated and encounters a recognizable chasing situation, most drivers will choose to hand over the driving control to Artificial Intelligence (AI); as for those who choose to use AI to fulfill prior driving decisions, the decision mostly chooses to reduce casualties, but the result of reducing casualties actually leads to harm to passersby who obey the traffic rules. Our findings indicate that ethical issues are inevitably involved in human-robot cooperation and conflict, and also make us aware of the implications of re-examining traffic regulations and human safety.