/
© 2026 RiffOn. All rights reserved.

Get your free personalized podcast brief

We scan new podcasts and send you the top 5 insights daily.

  1. Making Sense with Sam Harris
  2. #494 — A Coin Toss for the Future
#494 — A Coin Toss for the Future

#494 — A Coin Toss for the Future

Making Sense with Sam Harris · Sep 22, 2026

AI safety expert Ryan Greenblatt estimates a 50-60% chance of AI takeover, discussing the paradox of the current AI arms race amid high risks.

Recursive AI Self-Improvement Could Compress a Decade of Progress Into One Year

Once AI systems become proficient at AI R&D, they can trigger a recursive self-improvement loop. This process could radically accelerate progress, potentially achieving an amount of algorithmic advancement in a single year that previously took over a decade. This "intelligence explosion" could rapidly create wildly superhuman systems from a starting point of mere human-level competence.

#494 — A Coin Toss for the Future thumbnail

#494 — A Coin Toss for the Future

Making Sense with Sam Harris·2 months ago

AGI Won't Emerge at Human-Level Parity, But as a Composite of Superhuman Skills

The moment Artificial General Intelligence (AGI) is achieved, it won't be like creating a new human-level mind. Instead, it will be the integration of many existing narrow AI capabilities, each of which is already vastly superhuman (e.g., in math or logic). Therefore, the first "general" intelligence will immediately possess a powerful, non-human profile of extreme strengths from day one.

#494 — A Coin Toss for the Future thumbnail

#494 — A Coin Toss for the Future

Making Sense with Sam Harris·2 months ago

AI Safety Scientist Estimates a 50-60% Chance of Misaligned AI Takeover

Redwood Research's Chief Scientist Ryan Greenblatt quantifies existential risk not as a small tail risk, but as a coin toss. He believes on the current trajectory, there is a 50-60% probability that misaligned AIs will seize control, with a significant chance of human extinction following such an event.

#494 — A Coin Toss for the Future thumbnail

#494 — A Coin Toss for the Future

Making Sense with Sam Harris·2 months ago

AI Risk Skeptics Often Downplay Danger by Using a Weaker Definition of AGI

Many who are dismissive of AI alignment problems aren't denying superintelligence is possible. Instead, they often implicitly use a lower capability threshold for terms like "AGI." They may imagine a system that is good at math and coding but cannot automate more complex strategic tasks, thereby avoiding contemplation of a truly world-altering intelligence.

#494 — A Coin Toss for the Future thumbnail

#494 — A Coin Toss for the Future

Making Sense with Sam Harris·2 months ago

AI Labs Race Ahead Despite Risks, Believing They Are the Safest Actor

The paradox of why AI development accelerates despite stated existential risks is explained by a competitive dynamic. Each major lab, like OpenAI or Anthropic, believes it is the most responsible party to develop the technology first, fearing a less cautious competitor will win. This creates an arms race where slowing down is seen as ceding control to a more dangerous actor.

#494 — A Coin Toss for the Future thumbnail

#494 — A Coin Toss for the Future

Making Sense with Sam Harris·2 months ago