Browsing by Subject "Trolley Problem"

Now showing 1 - 1 of 1

Embedding Ethical Theory Into Autonomous Vehicles: Analyzing a Covergence of Ethics and Action Embedding Ethical Theory Into Autonomous Vehicles: Analyzing a Convergence of Ethics and Action
(2020-05) Atchley, Austin
The ability of autonomous vehicle systems to collect data and, independently of passion, make split-second decisions creates a newly emerging phenomenon: developers encode their ethical views into something that will (almost deterministically) enact them in reality. This is a departure from the traditional relationship between belief and action that is present in each decision made by a human, and, as artificial intelligence becomes more widespread, perhaps a majority of day-to-day decisions will be made without human unpredictability. Instead, decision-making will boil down to the application of ethical principles (either explicitly or implicitly) to a dataset. In creating ethical autonomous vehicles, we should address the problem using both top-down and bottom-up approaches to encoding ethical beliefs. The top-down approach uses high-level ethical values defined by the software creator to drive decision-making processes, and the bottom-up approach attempts to mimic human ethical conduct by using a pre-collected dataset of decisions made by real human drivers. Currently, many developers are not aware of the ethical principles they embed into their code, and by taking an explicit approach to ethical decision-making, we can encourage morally preferable decisions to be taken by autonomous vehicles. These issues are inherently interdisciplinary, and this thesis treats them as such, borrowing from both engineering and philosophical discourse. I argue that a similarly holistic perspective should be adopted by scholars working on the many-faceted topic of autonomous driving.