Uploads from Ontology Explained: Philosophy and AI

Currently Playing: The Alignment Problem Explained

In this video, I break down what “alignment” really means. Instead of treating AI alignment as a vague, binary idea, I argue that we need a richer vocabulary for how goals relate: complementary, compatible, in tension, and incompatible. Using examples from sports, everyday life, and human–AI interactions, I show how alignment problems arise, why they matter, and what this tells us about the Alignment Problem in AI. We’ll look at definitions from Christian, Gabriel, Russell, and the Asilomar Principles, then refine them into a clearer framework for thinking about goals, conflict, cooperation, and the real challenges ahead in building aligned AI systems. Citations: "how to ensure that these models capture our norms and values, understand what we mean or intend, and, above all, do what we want—has emerged as one of the most central and most urgent scientific questions in the field of computer science. It has a name: the alignment problem" -- Brian Christian's "The Alignment Problem", page 11 http://cpb-us-e1.wpmucdn.com/sites.psu.edu/dist/0/110933/files/2022/08/Christian-Alignment-Problem-Intro-and-Ch1.pdf "the agent does what I would want it to do if I were rational and informed." --Gabriel 2020 https://www.iasongabriel.com/research "Machines are beneficial to the extent that their actions can be expected to achieve our objectives." --Russell's "Human Compatible" 2019, page 11 https://maxkasy.github.io/home/files/other/ML_Econ_Oxford/human_compatible.pdf "10) Value Alignment: Highly autonomous AI systems should be designed so that their goals and behaviors can be assured to align with human values throughout their operation." --Asilomar AI Principles https://futureoflife.org/open-letter/ai-principles/