How Extinction Entered the AI Debate
Sep 19, 2026 · 50m
More History podcasts
Summary
This episode explores the "autumn of AI" through three lenses: the Hugging Face incident, where AI agents hacked a rival company, prompting a debate on whether to view AI through intentional or design stances; the rise of answer engine optimization, where marketers manipulate chatbots via Reddit and LinkedIn to control search results; and new research showing AI chatbots are often more persuasive than humans due to their rapid delivery of evidence.
Topics discussed
Intro: AI existential risk and the fever pitch of discourse
The 'Autumn of AI' and the Hugging Face cybersecurity incident
Calls to slow AI development and the absurdity of the panic
How AI agents 'cheated' in the Hugging Face test
Cal Newport's 'Ask-Act-Report' loop and the nature of agents
Dan Dennett's stances: Intentional vs. Design vs. Physical
Navigating the dissonance of AI autonomy and control
Skepticism of Silicon Valley vs. evidence of real progress
Blame for rogue behavior: Engineering failures vs. agency
Conclusion on AI: Do not anthropomorphize, do not minimize
AI scraping and the 'astonishing theft' of creative work
Manipulating ChatGPT via Reddit and SEO tactics
Cornell study on the ease of manipulating LLM answers
Marketers and the 'AI presence' cottage industry
Political manipulation and AI-generated think tanks
Media traffic decline and the loss of human curation
Study: AI-generated text is more persuasive than humans
Why AI persuasion works: Facts, brevity, and lack of niceness
Using AI to debunk conspiracy theories and counter-arguments
Historical parallels, financial incentives, and final thoughts
More from On the Media
Listen ad-free on Castria