Evidence based policy cause area report

Abstract

This article surveys a variety of foundational texts, recommended for understanding human-compatible artificial intelligence (HCAI) and its surrounding research fields. These texts are grouped into 8 main categories: theoretical background, introduction to AI/ML, prevailing methods in AI/ML, broad perspectives on HCAI, open technical problems, social science perspectives, cognitive science, and moral theory. Each category contains a list of textbooks, course materials, published and unpublished articles, news and magazine articles, and blog posts. The studies included in this work range from the technical challenge of designing robust and corrigible AI systems, to investigations into human oversight and interactive AI, and to deep dives into ethical and philosophical quandaries that accompany the development of advanced artificial intelligence. – AI-generated abstract.