ETHICS FOR ARTIFICIAL HISTORIANS

Abstract

Artificial historians do not need to have intentions to complete actions or to solve problems. Consequently, a revised approach to the ethics of history is needed. An approach to ethics for artificial historians can be proposed through the recognition of historiographical logic, which is a hybrid of modal, propositional, and erotetic (question-based) types. Looking to examples of texts produced by artificial and human historians, I argue that this hybrid historiographical logic is seen at play in what Jo Guldi has called “signal” (which can denote both the focus and interpretation of historians) and I call “healthy noise,” or metadiscursive question-begging and possible world generation. Together, the signal and noise of histories generate an ethical stance of openness toward the possibilities of new evidence from the past, new ways of interpreting that evidence, and new combinations of signal and noise. Recognizing the potential presence of this logic in modal-propositional-erotetic statements about the past in a wide range of texts, I argue for the recognition of historiographical logic as broadly useful for AI and specifically useful for the discipline. This historiographical and logical turn facilitates stepping beyond treating ethics as the generation of lists of principles that may fail in application and toward recognizing particular forms of logic as indicating ethically beneficent outcomes. This shift may facilitate the detection and deterrence of histories shaped by logics that indicate ethically maleficent outcomes.