With the advent of increasingly automated machinery in daily life, machines will increasingly have to make ethical decisions. This project aims to examine the underlying assumptions and problems in computation of ethical dilemmas. It is the thesis of this project that there exist decision procedures that are sufficiently extensionally equivalent to ethical theories for practical usage, and that these decision procedures are computable. This is done by examining the limits of computation and extensional systems as discussed by Alan Turing, Alonzo Church, and Kurt Gödel. Algorithms are then proposed, ¨ and their limitations exposed. Common weaknesses come from nonexistence of expected value such as where utility is predicted with a Cauchy-Lorentz random variable, or from inability to compare alternative situations such as Arrow's Impossibility Theorem. Despite these limitations, it is shown that it is still theoretically possible for computable decision procedures to be largely extensionally equivalent to ethical theories, specifically Utilitarianism and Kantian Deontological Ethics, despite limitations that mean true equivalence is impossible.
Hale, David Edward, "Ethics in Computing" (2019). Senior Independent Study Theses. Paper 8471.
Applied Ethics | Artificial Intelligence and Robotics | Databases and Information Systems | Logic and Foundations of Mathematics | Numerical Analysis and Computation | Other Applied Mathematics | Other Computer Sciences | Philosophy of Language | Philosophy of Mind | Probability | Statistical Models | Theory and Algorithms
Church-Turing Thesis, Utilitarianism, Utility, Measurement, Expected Value, Gödel's Incompleteness Theorem, Arrow's Impossibility Theorem, Ethics, Computation of Ethics, Kantianism, Neoclassical Economics, Extensional Equivalence, Intensionality, Theory of Mind, Theory of Language, Turing Machines, Weak AI, Strong AI
Bachelor of Arts
Senior Independent Study Thesis
© Copyright 2019 David Edward Hale