Intelligent Machines 840: Pudding Forks

Podcast: All TWiT.tv Shows (Audio)
Date: October 9, 2025
Host: Leo Laporte
Co-hosts: Paris Martineau, Jeff Jarvis
Theme: The current and future state of AI—from viral “AI slop” apps and the blurring line of digital reality, to industrial “bubbles,” data center economics, and pudding forks.

Episode Overview

In this wide-ranging discussion, Leo Laporte, Paris Martineau, and Jeff Jarvis cover the latest trends and controversies in artificial intelligence, from viral apps like OpenAI's Sora and their social implications, to deeper questions about AI’s impact on jobs, information integrity, data center economics, and tech business models. Alongside, the hosts touch on cultural phenomena—a German pudding-eating craze—and sprinkle in playful banter, personal anecdotes, and the enduring skepticism/optimism divide around tech.

Key Discussion Points & Insights

1. OpenAI's Sora and the Era of “AI Slop”

(09:41, 10:43, 11:05, 11:15, 13:19)

• Sora, OpenAI’s new viral app, is the number one app on iPhone, enabling users to create video “cameos” of themselves and others.
• Hosts experiment with Sora—posing for camera, using 2-digit codes, and witnessing deepfake videos of themselves. The tech impresses, but also produces “AI slop,” raising concerns about misuse and loss of control.
• “Do you get notification when somebody uses your image?” (13:19, Jeff) — Sora allows public/private settings but limited deletion controls.
• Concerns include identity misuse and plausible deniability: “If you see a video of me doing something awful, I could just say, ‘well, that’s AI.’” (14:30, Leo)

2. Deepfakes and the Erosion of Truth

(18:32, 19:19-21:09)

• The panel debates the growing challenge of verifying reality in an age of sophisticated AI-generated media, referencing a viral Kamala Harris clip as a case study.
• “We’re just not going to be able to believe anything we see anymore.” (20:09, Leo)
• Paris mourns that future historians won’t be able to trust video as records; Jeff analogizes the current mistrust to the early days of print and radio.

3. Data Center Economics & AI Business Models

(22:02-26:59, 50:26-54:17, 58:14-66:23)

• OpenAI’s scale is immense: ChatGPT reports 800 million weekly active users. Uncertainty persists about profitability—rumors that generating a Sora video may cost up to $5.
• “He’s like Uber. They’re losing money on every transaction. Right. But they’re building users.” (22:29, Leo)
• US tech sector bets big on AI, with 80% of this year’s stock market gains attributed to AI companies.
• Data centers now account for 4% of GDP (51:38, Jeff) and up to 92% of GDP growth in early 2025 is tied to information processing equipment/software.

4. AI “Industrial Bubble”—Jeff Bezos' Perspective

(29:01-32:23, 119:32)

• Leo shares Jeff Bezos' analogy of the current AI surge as an “industrial bubble,” contrasting it to destructive financial bubbles of the past.
• “AI is real and it is going to change every industry... it’s a horizontal enabling layer.” (30:09, Jeff, quoting Bezos)
• Even if many AI startups disappear, the fundamental technologies and infrastructure remain—a “transcontinental railroad” moment for tech.

5. Jobs and AI: Panic vs. Reality

(33:34-36:39, 39:47-42:26)

• Senate Democrats claim nearly 100 million jobs could be lost to AI in the next decade—a number Jeff and Paris find dubious.
• Some evidence suggests current AI has had “essentially zero impact on jobs” (34:29, Leo citing Yale).
• The hosts discuss which jobs might be most and least at risk, poking holes in categorical predictions (e.g., food service, stock movers, executive assistants).
• Sam Altman, CEO of OpenAI, claims even CEOs may one day be replaced by AI—“deflection” and “bubble” talk, the hosts suspect (39:55).

6. Cautionary Tales: Trust in AI Output

(47:02-50:19)

• Deloitte’s fiasco: They produced a report for the Australian government discovered to be full of fabricated references and invented quotes after using AI to generate content.
• “They used AI to write this thing. And there were fabricated quotes from a federal court judgment, references to non existent academic reference papers.” (47:58, Leo)
• Shortly after, Deloitte and Anthropic announced a partnership to build “AI solutions for regulated industries,” raising concerns about due diligence and fact-checking in the AI era.

7. The Social Impact of AI: The Case of NYC’s Friend Ads

(70:30-76:12)

• NYC subways are saturated with ads for “Friend,” an AI-powered “companion” pin; almost every ad is vandalized—“AI is not your friend.”
• “Don’t use AI to cure your loneliness. Reach out into the world.” (75:00, Paris, quoting a defaced ad)
• This groundswell of anti-AI sentiment in public spaces signals a broader skepticism and anxiety about digital relationships and surveillance.

8. AI Hype, Cynicism, and the Search for Meaning

(54:17, 56:14, 57:02, 66:23)

• The group discusses whether artificial general intelligence (AGI) is an attainable—or even meaningful—goal.
• “We should stop asking, ‘Is the machine intelligent?’ And ask, ‘What exactly does it do?’” (54:17, Leo quoting Shannon Valor).
• Ongoing tension between AI prophesies, naysayers, and the real utility/limitations of today’s models.

9. Business Models and the Future of AI Apps

(59:34-65:20)

• The hosts debate potential endgame business models—direct payment vs. advertising vs. AI as a wholesale technology sold to embedded platform/app providers.
• Leo analogizes the challenge to running a high-end sandwich shop: “The sandwich costs you $30 to make... but the good news is you’re selling every sandwich you can make.” (60:42)
• Jarvis predicts AI “models” will become commoditized and application “layers”—where users encounter AI—will determine monetization and value.

10. Playful & Cultural Asides

• German Pudding Forks Phenomenon (154:52-157:34): A viral, absurdist trend in Germany where crowds gather to eat pudding with forks—“pure absurdism” and a case study in how digital culture morphs into meme-like reality.
  • “People are meeting on up in Germany and Austria right now... each with their own pudding, each with their own fork...” (156:03)
• Personal Anecdotes & Tech Nostalgia: The gang reminisces about Tivo, VCRs, Marion Stokes’ archive, and the rise and fall of various platforms and gadgets.

Notable Quotes & Moments

• “No, no, no. That’s the point... this kind of nonsense muddies the water for the useful, sensible things about werewolves.” — Leo Laporte (07:27)
• “We’re just not going to be able to believe anything we see anymore.” — Leo Laporte (20:09)
• “When print came out, it was not trusted because the provenance was not clear and anybody could make it... they come up with new norms and institutions to figure this out.” — Jeff Jarvis (20:45)
• “AI is real and it is going to change every industry... it’s a horizontal enabling layer.” — Jeff Jarvis, quoting Jeff Bezos (30:09)
• “I mourn the idea that future historians won’t be able to look back on video records and know with any certainty whether or not those things happened.” — Paris Martineau (20:34)
• “Deloitte confirmed some footnotes and references were incorrect... They used AI to write this thing.” — Leo Laporte (47:58)
• “Sycophantic AI decreases pro social intentions and promotes dependence... people are drawn to AI that unquestionably validate—even though that validation risks eroding their judgment.” — Jeff Jarvis (99:22, 101:01)

Important Timestamps

| Timestamp | Segment / Topic | |-----------|-------------------------------------------------------------------| | 09:41 | OpenAI Sora app discussion, viral “AI slop” | | 13:19 | Sora: Image usage, notifications, and deletion | | 18:32 | Deepfakes: Kamala Harris MF clip, verifying authenticity | | 22:02 | AI economics: OpenAI usage, Sora video costs, money burn | | 29:01 | Jeff Bezos on “industrial bubble” in AI | | 33:34 | Discussion: Job losses, AI impact, skepticism about big numbers | | 47:02 | Deloitte/Australia: AI-generated report fiction | | 70:30 | NYC Subway Friend ads: Public backlash and vandalism | | 154:52 | Pudding with forks: German pudding-mit-Gabel phenomenon | | 99:22 | Sycophancy in AI: Academic findings, warnings |

Tone & Style

• Conversational, witty, and occasionally irreverent—the hosts riff on personal tech histories, mock AI failures and hype, but also challenge and scrutinize each other’s assumptions.
• The show is a blend of informed opinions, historical context (print, radio, Tivo, etc.), and honest trepidation about the social consequences of current trends.
• There’s a healthy skepticism but also a playful, sometimes absurdist, approach (especially re: pudding forks and German subcultures).

Closing Notes & Picks

• Podcast Pick (Paris): “The Necessary Conversation”—a podcast where left-leaning children and ultra-MAGA parents hash out politics weekly, “brave to do and brave to listen.”
• Trend Watch (Jeff): “Pudding mit Gabel”—Absurdist gatherings in Germany where crowds eat pudding with forks, a viral curiosity.
• Acknowledgements: Nod to journalist Glenn Fleishman’s upcoming surgery and support via his books.

Next Episode: Guest Jeffrey Quinnell from Nous Research, on emotionally intelligent, human-centric AI.

Summary for the Uninitiated

This episode is a rollicking guide to the weird, wild, and worrying present of AI: from the viral to the infrastructural, the playful to the profound, the economic to the existential. It offers a front-row seat to how experienced tech journalists and thinkers are parsing the rapidly shifting landscape—with all its slop, promise, panic, and pudding.