Podcast Summary: How Superintelligent AI Could Upend Work and Politics

Podcast: Plain English with Derek Thompson
Host: Derek Thompson
Guest: Anton Korinek, Economist at the University of Virginia
Episode Date: November 18, 2025

Episode Overview

This episode delves into the potentially profound economic, social, and political upheavals that could occur with the development of superintelligent AI, or Artificial General Intelligence (AGI). Host Derek Thompson and economist Anton Korinek discuss what might happen if AGI arrives within the next few years, focusing less on catastrophic or doomsday scenarios and more on realistic, even positive, but deeply disruptive consequences for work, prices, productivity, and politics.

Key Discussion Points & Insights

The Frankenstein Parallel & Setting the Stage

• Thompson opens by comparing Mary Shelley's Frankenstein to today’s anxieties over AGI. The perennial myth of humanity creating something it cannot control (“a modern Prometheus”) reflects our era's questions about superintelligent AI.
  • "[Frankenstein] has been made and remade... because it answers a question that... we can't stop asking ourselves: are we human beings capable of creating an invention that rises up to destroy us?" (03:45)

Is the AI Boom a Speculative Bubble? (09:20)

• Anton Korinek: AI investment today feels like the dot-com or housing booms—full of speculative frenzy—but might actually yield the most transformative technology in history.
• Why so impactful?
  • If AGI comes to fruition—machines capable of performing almost all economically valuable work—it's akin to “a country of geniuses in a data center” (10:20).

Why Believe AGI Hype? (11:37)

• Three reasons AGI might really happen soon:

  1. Scaling Trends: Continued improvements as more computational power is applied.
  2. Firsthand Evidence: Rapid, surprising capability improvements in AI tools over just the past 2 years.
  3. Biological Proof-of-Concept: The brain itself is proof that neural networks (biological or artificial) can be made generally intelligent.
  • “...if you catch them [AI company leaders] in private, they are probably going to say as well that this is a bet. There is no certainty. But...we're on a curve...” — Korinek (11:37)

Are Today’s AI Models Already AGI? (17:02)

• Thompson observes that today’s systems already outperform any single human in many white-collar domains. So why aren’t we already calling this AGI?
• Korinek: We constantly move the goalposts (“shifting expectations”) for what counts as AGI. Current models lack abilities like dynamic, lasting learning and interacting with the world, but their raw capabilities are “already very powerful” and would yield significant economic growth if just today’s tech was widely deployed (18:06).

The Arrival: "Quiet" vs "Loud" AGI (20:00)

• Scenario 1 – Quiet AGI: AGI arrives, but it takes years to notice its impact; no instant headlines or productivity surge.
• Scenario 2 – Loud AGI: AGI causes immediate, dramatic breakthroughs or disruptions, e.g. curing cancer overnight or hacking a world infrastructure system.
• Korinek: A middle ground (“not super loud, but impactful”) would be ideal—swiftly using capabilities for productive purposes without hype-fueled backlash (21:53).

Will AGI Lead to Rapid Problem Solving? (25:32)

• Will AGI immediately solve society’s biggest issues?
• Korinek: Even with a “data center full of genius economists or scientists,” physical and process bottlenecks (like drug testing and regulation) remain. Intelligence alone isn’t a panacea; many problems are "really, really hard" (27:59).
• "[Bottlenecks] make progress faster, but it doesn’t make every problem solved by itself." — Korinek (27:05)

The Recursive Loop & First Effects (29:55)

• Software that writes better software: recursive self-improvement and its disruptive potential.

• Likely short-term impact: First, computer programming jobs could be automated away—by the very AGI those workers built.

  • “Doesn’t that suggest...the first effect of AGI will be to disemploy the software programmers who built the AGI?” — Thompson (29:55)

Productivity Surges and Strange Price Effects (33:59)

• Baumol’s Cost Disease: Productivity advances in some sectors—like software and research—while others, like personal care, remain dependent on humans and become much more expensive.

  • “Whatever you can’t automate becomes the bottleneck.” — Korinek (36:06)

• Speculation: If cognitive work becomes "dirt cheap" but haircuts, food service, and healthcare remain manual, prices for these rise sharply (36:06).

Will AGI Replace or Empower White-Collar Workers? (38:28)

• Thompson provides a counter-example: Despite predictions, radiologists became more numerous and better paid after AI arrived—they use AI as a powerful tool.
• Will LLMs be the “Excel” of the future—empowering rather than replacing workers?
• Korinek: For a true AGI, “there’s suddenly nothing left that only the human could do.” The Excel analogy only goes so far; AGI, in theory, can do every task humans can, even if that arrives gradually (40:42).

The Physical Bottleneck: Robotics (42:53)

• Many jobs—caregiving, food service, construction—require physical dexterity and presence.
• Korinek: Large progress underway (“significant advances in robotics”), but generally capable robots will probably lag behind AGI by a few years. Once AGI arrives, rapid robotics development is likely (45:35).

Everyday Changes from a Robotics Revolution (46:17)

• Picture for the 2030s: Fast food restaurants and many similar services could be entirely automated, with few or no humans involved “within a decade or so.”
• McDonald's with robot workers as the norm (46:17).

Will AGI Permanently Destroy Human Jobs? (48:21)

• Three possible long-term “human job” domains:

  1. Spiritual activities (therapy, religion, etc.)
  2. Performative activities (sports, arts)
  3. Overseeing and managing AIs

• Temporary vs Permanent Joblessness?

  • If disruption is slow and we value human involvement, we might avoid “Great Depression” levels of mass unemployment.
  • If change is fast, we could see 20% or higher unemployment absent major government intervention (48:21).

Political & Social Backlash (51:26)

• Thompson: “This is where my brain breaks”—if huge swaths of the workforce are disemployed, political pressure will likely force a halt or reversal of AGI deployment.

• Examples from history (Great Depression and anti-tech backlash).

• What to expect: Strong competitive forces will drive automation and profit, but failing to help “the losers” will provoke severe democratic backlash.

  • “...nobody wants to be the loser that has to be taken care of...people are the heroes of their own lives.” — Thompson (54:05)

Policy Solutions: No American Left Behind (55:55)

• Korinek: There must be large-scale social support and wealth redistribution if AGI dramatically increases societal wealth without equally distributed benefits.
  • “We need something like a ‘No American Left Behind’ program to make sure those challenges...are addressed...It's the only way in a functioning democracy to have disruptive technological progress without...major backlash.” (55:55)

Notable Quotes & Memorable Moments

• “If we have artificial general intelligence...it would utterly transform our world.” — Korinek (10:01)
• “If you had shown anybody what today's LLMs can do ten years ago, they would have probably said, ‘Yes, this is AGI.’” — Korinek (18:06)
• “Whatever you can't automate becomes the bottleneck.” — Korinek on cost disease (36:06)
• “It’s plausible that we’ll have a transition...without, let’s say, Great Depression levels of disruption. I hope that our world is going to look that way.” — Korinek (48:21)
• “Nobody wants to be the loser that has to be taken care of...they want to be thought of as winners, and they aspire to be winners.” — Thompson (54:05)

Timestamps for Major Segments

• 03:45: Frankenstein, technology, and myth
• 09:20: Is AI hype a bubble?
• 11:37: Signs we’re on track for AGI
• 17:02: Are today's AI systems already AGI?
• 20:00: Quiet vs. Loud AGI—how might it “arrive”?
• 25:32: What would AGI really change immediately?
• 29:55: Recursive self-improvement and automation of programming jobs
• 33:59: Productivity leaps, Baumol’s cost disease, and strange inflation
• 38:28: Will LLMs empower or obsolete workers?
• 42:53: The robotics revolution and physical work
• 46:17: A robotized daily life
• 48:21: How many jobs might vanish, and for how long?
• 51:26: The “brain break”—politics, backlash, and democracy
• 55:55: Policy ideas—no one left behind

Tone & Language

The conversation is accessible but intellectually rigorous, alternating between pragmatic speculation, dry humor, and a clear-eyed assessment of economic and political risks. Thompson’s skepticism and curiosity match Korinek’s analytic optimism and realism, resulting in a highly engaging exploration of AGI’s possible futures.

Final Takeaways

This episode offers a nuanced overview of the near- and medium-term future if AGI arrives:

• Incredible technological achievement could yield vast economic gains.
• Major disruptions to existing labor markets and possibly societal structure are likely.
• Without proactive policy, democratic societies may experience severe backlash.
• Despite the sci-fi premise, the real upheavals will be intensely human: jobs, self-worth, politics, and the meaning of work.

For anyone thinking about how AI might genuinely change the world—economically, socially, or politically—this episode is a must-listen.