Topic: Superintelligence

3 chapters across the catalog

A Dog A Day
Episode 1842 49:07 - 58:02

1842: A Dog A Day

Microsoft AI CEO, Future of White Collar Automation

The CEO of Microsoft AI predicts that most professional white-collar tasks, including law and accounting, will be fully automated within 12 to 18 months. Meanwhile, Apple is reportedly struggling to integrate similar AI capabilities into Siri, with internal testing revealing accuracy and latency issues that may delay features until 2026 or 2027. The shift in software engineering is described as moving from code production to strategic architecture.

Battle Rhythm
Episode 1824 12:19 - 23:18

1824: Battle Rhythm

Nate Soares, Superhuman AI Extinction Risks

Nate Soares, president of the Machine Intelligence Research Institute, appeared on ABC with George Stephanopoulos to discuss his book regarding the existential threats of superhuman AI. Soares argues that current AI development techniques could lead to human extinction if an indifferent superintelligence is created. The discussion touches on the lack of federal regulation and the appointment of David Sachs as AI czar.

Trusted Flaggers
Episode 1544 1:21:34 - 1:26:39

1544: Trusted Flaggers

Eliezer Yudkowsky and the AI Open Source Debate

The hosts critique an interview with AI researcher Eliezer Yudkowsky, who argues against open-sourcing powerful AI models like GPT-4 due to existential risks. They compare the current alarmism surrounding "superintelligence" to the rhetoric of the climate change movement, suggesting it is a new form of hype designed to generate funding and international junkets. The discussion dismisses the idea that AI will "ignite the atmosphere" as hyperbolic nonsense.