Intelligent Machines (Audio)

Podcast: Intelligent Machines (Audio)
Host: Leo Laporte
Co-hosts: Paris Martineau, Jeff Jarvis
Special Guest: Jimmy Wales (Founder of Wikipedia)
Episode: 846: "Chivelord - From Leather-Bound to Cloud Powered"
Release Date: November 20, 2025

Episode Overview

This episode welcomes Jimmy Wales, visionary founder of Wikipedia, to discuss his new book The Seven Rules of Trust, the evolution of knowledge from print encyclopedias to globally accessible wikis, and the challenge of trust in the age of AI and digital slop. The conversation covers the human and technical foundations of Wikipedia, how trust and community operate online, the peril/possibility of AI for collective knowledge, and the enduring need for purpose and "niceness" in internet communities.

Key Discussion Points and Insights

1. Wikipedia: From leather-bound books to cloud-powered communities

[02:27-04:30]

• Jimmy shares his early frustration as a new parent (in 2006) seeking accurate medical information online, leading to the founding of Wikipedia.
• Wikipedia has grown to 7 million English-language articles, millions of contributors, and 300+ languages ("93 times bigger than Britannica" – Leo Laporte, [03:42]).
• Reflections on encyclopedias as performative middle-class furniture vs. living knowledge.

Memorable Quote:
"We always understood we were selling both a knowledge product and furniture." – Jimmy Wales, [04:30]

2. Founding the Wiki Model & Community Evolution

[07:03-12:04]

• Early experience: Nupedia was slow, academic, and exclusive; wikis brought instant collaboration and ease; credit to Ward Cunningham.
• Wales recalls the "leap of faith" releasing editorship to the crowd: "Does it really matter? It's really the work that matters, not the qualifications." [10:04]
• Technical early days: no passwords; anyone could impersonate anyone.

Memorable Moment:
"You could create a user account, but there was no concept of passwords ... So you could give yourself a name, but anybody else could pretend to be you!" – Jimmy Wales, [11:11]

3. Building and Maintaining Trust Online

[13:33-19:07]

• Trust is more than public perception—starts with trusting contributors.
• Early anxiety over vandalism, then discovery of the "dark energy" of responsible users worldwide.
• Cites Frances Frei’s ‘trust triangle’ (authenticity, empathy, logic) as a powerful framework for online community integrity. [15:15-17:04]
• Empathy is missing from much of today's online world—leads to polarization, misunderstanding, and brittle communities.

Memorable Quote:
"If you're not having any empathy whatsoever, then the interaction is cold and not helpful." – Jimmy Wales, [16:00]

4. Trust, Scale, and Identity on the Internet

[19:07-24:05]

• Wikipedia's reputation system is built on persistent, consistent identity (not necessarily real names).
• Social platforms like Twitter lack these structures, resulting in atomized, uncivil engagement.
• "If you post almost anything on Twitter ... some absolute random sort of sends you an angry message." – Wales, [20:36]
• Rule #5 in Wales’s book: "Your mother was right. Be nice, be civil, be respectful." [21:24]
• Strong, clear, positive purpose (like Wikipedia's encyclopedia mission) is key—something most social and AI companies lack.

5. Wikipedia’s Non-profit Model and Commitment to Purpose

[27:08-29:20]

• Decision to reject advertising for Wikipedia:
  • Not about money, but artistic/aesthetic purity ("It's like a temple for the mind" – Wales, [27:47])
  • Most fundraising comes from small donors (avg. just over $10).
  • "[Organizations] ultimately do follow the money... our funding model means nobody cares what you read, only that you value Wikipedia enough to donate." [27:47-29:10]

6. Trust and Artificial Intelligence (AI)

[29:31-38:20]

• Growing trust crisis with LLMs: Wikipedia’s content is widely used in AI training, yet AI is prone to "hallucination" (confidently making up facts).
• Internal experiments with using AI for Wikipedia content curation have been met with skepticism:
  • AI is too eager to answer, too reluctant to say "I don't know."
  • Casual users may not spot plausible-sounding errors; caution required for credible knowledge bases.
• Anecdote: Wales tests new AIs by asking about his wife, Kate Garvey—the models invent detailed, plausible but false stories ([32:04-34:42]).

Memorable Story:
"My wife said, ‘That could have happened. We could have easily been at a party ... it's completely plausible but completely false.’" – Jimmy Wales, [33:30]

7. Practical, Human-centered Uses of AI for Wikipedia

[35:53-38:20]

• Appropriate role for AI: help with translation, language refinement, and batch tasks (finding dead links, suggesting sources).
• Human judgment remains essential for accuracy and context; AI should handle the “boring bits.”
• AI is surprisingly creative, less rote/mechanical than expected—“Turns out it's really bad at regurgitating facts and it's quite good at brainstorming...” – Wales, [37:47]

8. Wikipedia vs. Grokopedia/AI-generated Knowledge| Neutrality and Hallucination

[38:20-42:01]

• Wales hasn’t studied Grokopedia (Elon Musk’s Wikipedia competitor) deeply but is skeptical:
  • Alignment with creators’ political views is a risk.
  • "AI is not good enough to do this ... it's going to make stuff up." [38:50]
  • Early research shows Grokopedia articles are longer, but less original and less well-sourced ([40:55]).

Funny Moment:
“Tell it to use bigly words, lots of hard words, that'll show that it's really smart.” – Jimmy Wales, [41:39]

9. Jimmy Wales, Retirement, and AI Experiments at Home

[42:14-46:10]

• Wales is a lifelong volunteer, never an employee at Wikimedia.
• New personal project: creating a home AI assistant that mimics the ghost of a 19th-century Admiral who lived in his house. (Playful, contextually aware, local LLM use).
• Strong endorsement for local, on-device language models and the need for powerful local computing for AI in near future ([44:06-46:34]).

10. The Future of AI—Specialization, Local Models, and Research Directions

[93:43-98:38]

• Discussion of specialized vs. general AIs:
  • Anticipates a future where little, purpose-specific AIs ("agents") proliferate.
  • AGI (General/Universal Intelligence) may not emerge from LLMs alone; new research is vital.

Notable Quotes & Moments

• On trust:
  “Building trust is very practical... It actually works.” – Jimmy Wales, [21:31]
• On empathy:
  “That lack of empathy, that lack of saying—why are people so upset?—is where it breaks down.” – Jimmy Wales, [17:04]
• On refusing ads:
  “I think Wikipedia is very nice. It's like a temple for the mind and I enjoy it.” – Jimmy Wales, [27:47]
• On AI hallucination:
  “One of the problems with hallucinations is they tend to be quite plausible...it's completely plausible but completely false.” – Jimmy Wales, [32:49]
• On Wikipedia’s model:
  “We’re here to build an encyclopedia. Even the talk page...it’s like, ‘How do we improve the article?’ That purpose makes our self-management so much easier.” – Jimmy Wales, [24:27]
• On future of home AI:
  “For the first time in forever...I could use a really powerful computer...the era...where a fast computer is meaningful again.” – Jimmy Wales, [45:07]

Important Timestamps

• [02:27-04:30] – Wikipedia’s impact vs. Encyclopaedia Britannica
• [07:03-12:04] – From Nupedia to Wikipedia: embracing openness
• [13:33-19:07] – Building trust; Frances Frei’s Triangle; empathy and polarization
• [20:36-21:31] – The power (and weakness) of persistent identity online
• [27:08-29:20] – Wikipedia’s non-profit model, fundraising, and value
• [31:13-35:53] – Challenges of using AI for fact-checking and content
• [38:20-42:01] – Grokopedia v. Wikipedia: bias, AI slop, research findings
• [42:14-46:34] – Wales’ AI “Ghost Admiral” home assistant; rise of local LLMs

Additional Segments (Post-Interview & Broader AI News)

• [53:19-63:13] – Gemini 3 released, ChatGPT-5.1, Grok 4.1 updates; testing and impressions
• [83:51-85:51] – AI slop & content detection discourse (Kagi’s Slop Stop, challenge of large-scale detection)
• [91:11-94:00] – Local AI v. Cloud AI; need for on-device capability for privacy and speed
• [107:22-109:42] – Breaking news: White House considering executive order to preempt state AI laws
• [110:02-113:00] – Nvidia profits and the financial ecosystem around AI hardware and research
• [119:30-120:54] – Discussion of member-supported journalism and podcasts as alternatives to advertising

Tone and Language

The conversation throughout is lively, collegial, thoughtful, with ample humor (“temple of the mind,” “furniture,” “bigly words,” the Bamboo Metaphor debate, etc.). Wales is measured, self-deprecating, and passionate about trust, openness, and purpose. The hosts are both reverent and irreverent, probing technical and philosophical issues with curiosity and wit.

Additional Noteworthy Quotes

• On the AI hallucination problem:
  “This could have happened ... but it didn’t. It’s as if [AI] is creating my digital twin!” – Jimmy Wales, [33:31]
• On Wikipedia’s purity:
  “Organizations ultimately do follow the money ... our business model means nobody ever thinks about monetizing what you read.” – Wales, [27:47-29:10]
• On AI vs. humans:
  “The best use of any technology is the boring bits that don’t really require a human.” – Wales, [37:53]

Summary

This episode is a must-listen for anyone interested in the evolution of online knowledge, community governance, the foundations of trust in open systems, and the high-stakes challenge of AI in the 21st century. Jimmy Wales’s rare blend of technical know-how, community insight, and ethical clarity offers practical guidance for building things that last—whether they be wikis, charities, or AI-powered futures.