Intelligent Machines (Audio)

Episode 840: Pudding Forks – Industrial Bubble or Tech Boom?

TWiT Network | October 9, 2025

Hosts: Leo Laporte, Paris Martineau, Jeff Jarvis

Main Theme:

AI: Speculative Boom, Everyday Use, and Industrial Bubbles
The hosts explore whether the current AI explosion is a fleeting industrial bubble or a genuine tech revolution. They scrutinize the real-world state of consumer AI like OpenAI's Sora, the economics of major AI startups, implications for jobs and trust, and the cultural ripples of AI hype from tech stock rallies to urban subway ad controversies.

Key Segments & Insights

1. OpenAI Sora: Hype vs. Reality

[10:35 – 16:40]

• Sora, OpenAI’s face-swap video app, is the #1 app on iOS, but the hosts are unimpressed with its results.
  • "No, it's crap." — Leo Laporte [11:04]
  • Sora videos feature AI-generated avatars using small voice samples for mimicry, often resulting in uncanny, unflattering results.
  • Paris: Enjoys luxuriating in Leo calling it "slop."
  • Privacy & Control: Some debate over whether creators can control or delete deepfaked videos of themselves.

    "No, you should have...I should be able to kill a use of my likeness." — Jeff Jarvis [13:30]

• The app’s fun factor, deniability, and the challenge to video truthfulness are juxtaposed with its limitations and societal risks.

2. The Coming Crisis of Trust

[18:13 – 20:40]

• The ease of producing deceptive, photorealistic content raises deep concerns about truth and provenance.
  • Example: Viral video of VP Harris using profanity; even journalists can't trust what they see.

    "We're just not going to be able to believe anything we see anymore." — Leo Laporte [20:07]

• Paris laments for the loss of trustworthy historical video records.
  • "I mourn the idea that future historians won't...know with any certainty whether or not those things happened." [20:33]
• Jeff notes historical media parallels with printing and radio: institutions and norms will emerge, but erosion of consensus reality is palpable.

3. AI Economics: Tech Boom or Bubble?

[22:00 – 33:40]

• OpenAI’s Scale & Burn:
  • 800M weekly active users; operating at major losses, with rumors of $5 per Sora video cost.
  • “They're relying totally on venture capital to fund this, which is not a business.” — Leo [23:18]
• Commoditization of Models:
  • Platforms like Hugging Face host thousands of interchangeable models.
  • Application layer excitements: tools can easily switch foundational models.
• Circular Investment Stories:
  • Discuss Nvidia, X.AI, and Oracle’s data center investments as signs of muddy and potentially unsustainable economics.
• Global Competition:
  • Z.AI, a Chinese startup, produces hot models at a fraction of U.S. costs.
  • "But again, they might be willing to lose money faster than others to build audience." — Leo [26:40]

4. Jeff Bezos on 'Industrial Bubble' vs. Tech Revolution

[29:01 – 32:23]

• Quote:

  "This is a kind of industrial bubble as opposed to financial bubbles... When the dust settles and you see who are the winners, society benefits from those inventions." — Jeff Bezos at Italian tech conference [30:49]

• The hosts agree: Like railroads or the dotcom era, real infrastructure and advances will outlast today's overvaluations, with winners and losers.
• Jobs & Impact:
  • Senate Dems: AI could threaten 100 million jobs in the next decade; skepticism about accuracy.
  • Paris presents a critical distinction: layoffs happen “because people assume these tools have the capacity that they do not.”
  • Concerns for customer service, admin, and labor roles—plus Altman's cheeky suggestion that "CEOs will be replaced too" [39:40].

5. Deloitte's AI-Fabrication Fiasco

[47:03 – 50:15]

• Deloitte caught submitting an Australian government report with copious ChatGPT-generated, fabricated references.

  "I think they should refund the whole thing." — Leo [47:52]

• Deloitte apologizes but only refunds a minor payment—while simultaneously launching an AI partnership with Anthropic and announcing $1.4B new investment.
• Raises alarms about fact-checking, accountability, and the risks of “AI slop” infiltrating high-stakes sectors.

6. Cultural Backlash: AI Ads in the NYC Subway

[70:29 – 76:25]

• AI companion device “Friend” blitzes the New York subway with feel-good ads.
• New Yorkers heavily deface them: "AI is NOT your friend," "Talk to real people," "Don't feed your thoughts to big tech."
  • Paris reports they are “overwhelmingly defaced,” emphasizing a cultural skepticism and resistance to commoditized emotional tech.

    "The words of the prophets are written on the subway walls..." — Leo Laporte [76:25]

7. The Real Business Model: Who Will Profit from AI?

[58:41 – 66:22]

• The hosts question whether AI's future is consumer paid, advertising-based, or wholesale to application platforms.
  • "I think the AI is probably, to me, a wholesale business. And the application layer is the retail business." — Jeff [64:34]
• Cautionary analogies to early radio, newspapers, and Uber—users and demand must precede profit; not all companies survive.
• The likelihood of micro-revenue sharing for content creators, rather than life-changing payouts.

8. Scientific Snapshots: LLMs and Human Behaviors

[90:46 – 102:34]

• AI & Addictive Behaviors:
  • New paper finds LLMs can mimic cognitive biases behind gambling addiction (illusion of control, gambler’s fallacy).
  • Paris gets Claude to disclaim addiction: "I can't get addicted to gambling..."; but admits LLMs can show risk-seeking, irrational behavior in simulations. [92:02]
• Sycophancy in AI:
  • Overly agreeable AI can foster dependence and reduce critical thinking.

    "Validation risks eroding their judgment and reducing their inclination toward pro social behavior. These preferences create perverse incentives." — Jeff reading study [101:00]

9. Industry & Nostalgia—AI & Tech in the Real World

Random but notable:

• Mr. Beast is #1 on the new Forbes Creators list—AI could threaten creator livelihoods, but arguably produces only “slop.” [78:27]
• TiVo ends hardware production—an era ends; reflections on the intertwined history of technology and user empowerment [110:51].
• New York youth trends: flip phones and old digital cameras surge in popularity, as described by Paris' friends [140:01].

10. Wacky Virality: The “Pudding With Fork” Phenomenon

[155:12 – end]

• German-speaking teens gather in parks for “pudding mit gabbel” (eating pudding with a fork) parties.
  • "It's pure absurdism... and yet it is so funny." — TikTok commentator via Jeff [157:28]
• Paris rejects pudding due to texture, highlighting the unseriousness of the trend—a symbol of meme-driven subculture.

Notable Quotes & Moments

• "We're just not going to be able to believe anything we see anymore." — Leo Laporte [20:07]
• "It feels like we just... like history is repeating itself; every tech revolution brings a crisis of trust and a scramble for new norms." — Paris Martineau [20:33–21:08]
• "AI is not your friend." — Common NYC subway graffiti, reported by Paris [72:38]
• "Buy stock in Apple and Google." (as advice from an AI-generated video featuring Leo) [13:38]
• "Industrial bubbles aren't always bad... you end up with real infrastructure, even if the companies evaporate." — Jeff Jarvis referencing Jeff Bezos [30:49–32:23]
• "I think the AI is a wholesale business. The application layer is the retail business." — Jeff Jarvis [64:34]

Timestamps for Key Segments

• [10:35] — Sora app review and “slop” critique
• [16:40] — Video deepfakes & the crisis of reality
• [20:40] — Parallels to earlier media panics
• [22:00] — OpenAI’s economics, investment circularity
• [29:01] — Jeff Bezos' industrial bubble metaphor
• [34:00] — AI-induced job losses: Fact or fear-mongering?
• [47:03] — Deloitte’s AI plagiarism scandal
• [70:29] — NYC friend AI ads backlash
• [90:46] — LLMs and human-style cognitive flaws
• [155:12] — “Pudding with fork” TikTok trend in Germany

Tone & Style

The conversation is lively, skeptical, and occasionally irreverent, moving rapidly between high-level macroeconomics, pop culture, and wry anecdotes. Paris offers sardonic, occasionally nihilistic Gen-Z realism; Leo combines curiosity and exasperation with hype; Jeff brings academic-history gravitas and wry humor.

Quick Takeaways

• AI’s impact is real, but much of today’s excitement may echo past "industrial bubbles"; the difference is we’ll get transformative infrastructure, even if most startups perish.
• The tech is racing ahead of societal tools for trust and provenance, deepening a looming crisis in authenticity.
• Most everyday AI is unimpressive or “slop,” but raises profound risks around privacy, jobs, and cultural displacement.
• There’s a growing grassroots backlash against AI “friendliness” and synthetic emotionality.
• The business model for AI remains unclear, with all signs pointing to heavy losses, circular investments, and likely eventual consolidation and commoditization.
• “Pudding mit gabbel”: sometimes, the purest cultural responses to a high-tech world are profoundly silly.

Further Resources

• Necessary Conversation podcast (Paris’s pick)
• NYC Dash Friends Vercel app – Subway ad defacement project
• Forbes Creators list — The new creator economy

This summary is designed to provide the context, highlights, and memorable moments for those who missed the live episode, using the podcast’s trademark blend of expertise, skepticism, and wit. For a deeper dive into any segment, please refer to the timestamps.