Show Notes for Episode 3
Why Machines Will Never Rule the World, and Why Scientific Terms Sound More Convincing in German (with Jobst Landgrebe)
Listen to Ep 3: [Spotify] [Apple podcasts] [YouTube]
Our first guest is Jobst Landgrebe, co-author (with Barry Smith) of Why Machines Will Never Rule the World: Artificial Intelligence Without Fear. He complicates our “no experts” premise immediately: many AI engineers are real technical experts, he argues—but often naïve interpreters of what AI is.
Landgrebe’s core claim: minds and living systems are complex in a thermodynamic sense, shaped by history and irreversibility. LLMs can absorb surface regularities and generate persuasive language, but they don’t grasp meaning in open contexts—so they sound intelligent while remaining fundamentally non-understanding.
Yuri pushes back with an approximation critique: “close enough” could still equal or surpass human capabilities. Airplanes don’t fly like birds, yet they outperform birds—so why couldn’t AI become functionally superior to humans in many domains without “real” understanding?
We also dig into:
non-ergodicity and why real life doesn’t repeat cleanly
why scaling may not lead to open-world intelligence
AI in biomedicine and software development: where it works, where it fails catastrophically
the real dangers: surveillance, propaganda, and AI’s military applications
a tense moment on RFK Jr. and public trust