Can we account for meaning and value in terms of thermodynamics and information theory? Complex systems thinker Artemy Kolchinsky thinks so. He joins me to explain how we can make the idea of “meaning” tractable in rigorous formal terms by filling in what’s been missing from classic information theory about what information is intrinsically meaningful for an entity. We discuss his seminal co-authored paper with David Wolpert on semantic information, then explore connections to John Vervaeke’s relevance realization model, learning, and AI. We also consider how this framework relates to evolutionary theory and the challenges of scaling it to measure meaning across various levels of complexity.
0:00 Introduction0:56 What is Meaning in Information Theoretic Terms?12:17 Measuring Meaning18:14 Meaningful Information and Relevance Realization26:33 Learning and Viability40:29 Meaning and AI Alignment48:58 Studying How Meaning Scales54:11 Semantic Information, Evolution, and Normativity1:05:23 Operationalizing Semantic Information Theory1:14:18 Conclusion
To hear more, visit
brendangrahamdempsey.substack.com
