What Would the AI Apocalypse Look Like? We Went Looking for Answers
Summary
Wired’s Uncanny Valley dissects the escalating AI safety debate by breaking down specific doomsday scenarios, including autonomous hacking, bioweapon creation, and the "paperclip" problem of uncontrolled agents. The hosts analyze conflicting statements from tech leaders like Sam Altman and Dario Amodei, questioning whether their calls for "pacing the frontier" are genuine safety measures or marketing tactics. They also explore the political landscape, noting how the Trump administration’s resistance to regulation and antitrust concerns make it difficult for AI companies to voluntarily slow …
Topics discussed
Sponsors: D&B, Serval, and Toyota
Intro: The AI safety debate and today's agenda
AI researchers resign and call for a slowdown
Sam Altman's comments on AI risks and control
Debate: Is the safety concern a marketing ploy?
Dario Amodei's three-step plan for AI safety
Critiques of the international regulation approach
Jensen Huang: Market forces vs. new regulations
The problem of unelected tech leaders dictating policy
Defining the three main AI doomsday scenarios
Nuclear codes and the War Games scenario
AI-triggered false alarms and geopolitical tensions
Current administration's response to AI threats
AI attacks on water and power infrastructure
AI-assisted bioweapons creation and deployment
AI escaping control: The paperclip problem
Current harms vs. speculative future risks
Recursive self-improvement and the 'why not just stop' argument
US political response: Trump, Sanders, and Bannon
Antitrust risks and the China competition angle
Wired/Tired: App launch and AI industry lingo
Credits and final sponsor reads
Listen ad-free on Castria