Intelligent Machines 854: Welcome to the Pit
Podcast: All TWiT.tv Shows (Audio)
Host: Leo Laporte
Co-hosts: Paris Martineau, Jeff Jarvis
Guest: Thomas Haigh
Date: January 22, 2026
Episode Overview
This episode centers on the history and branding of Artificial Intelligence (AI) as explored in historian Thomas Haigh’s soon-to-be-published book. The hosts dive into Haigh’s research, which reframes AI as a shifting brand rather than a linear technological discipline, challenging the commonly accepted stories of AI "winters" and booms. The conversation also surveys the evolution of AI technology, funding, public perception, and the immense cultural and economic momentum currently propelling AI forward. There’s a personal touch, with Jeff Jarvis podcasting from his hospital bed, plus some spirited banter about AI’s societal implications and industry gossip.
Key Discussion Points and Insights
1. The AI "Brand": What is AI, and Who Gets to Define It?
Guest: Thomas Haigh (Historian of Computing)
- Origin of the Term “AI”:
- John McCarthy coined “artificial intelligence” in 1955 for a Rockefeller Foundation grant proposal, attaching a catchy name to ongoing research and giving rise to the first famous AI conference at Dartmouth (05:46).
- AI as a Brand:
- Haigh’s thesis: The “AI” label is a brand as much as a discipline. It’s been a marketing tool, with shifting definitions and ambitions, unlike more stable computing fields like databases or graphics (06:07).
- Quote: “You can consider any academic discipline through the lens of being a brand... but I think within computer science... artificial intelligence has had a lot more historical change...” (07:03).
- Early Conceptions vs. Today:
- 1950s-60s: AI meant programming computers to do things only humans could do—no strict definition of “intelligence.”
- Over time, specific approaches (symbolic logic, neural nets, expert systems) alternated as focal points for “AI,” with neural networks even temporarily excluded from the “AI” brand and called things like “pattern recognition” or “machine learning” (10:14).
2. Re-examining AI Winters and Industry Narratives
-
Debunking the Multiple AI Winters Myth:
- Contrary to popular belief, Haigh argues there was only one true AI winter (the 1990s)—not a series (12:12).
- 1970s "winter" actually saw international association growth and research expansion, just not in a few privileged, primarily US-based elite labs which dominated the narrative (13:11).
- Quote: “A very specific historical perspective from a handful of elite lab leaders... warped our understanding of what was going on in AI in the ‘70s.” (15:22)
-
Role of Funding and Cold War Context:
- AI boomed or waned largely based on shifts in government funding, especially from ARPA/DARPA, influenced by Cold War anxieties and international tech competition (25:03).
- Recommendations that the “brand” shifts correspond less with actual progress and more with how well the field attracted political or commercial attention.
3. From AI Dreams to Commercial Realities
-
Evolution from General Intelligence to “Expert Systems”:
- When general AI ambitions hit obstacles, the field shifted toward narrower, economically driven expert systems (16:51).
- These systems (e.g., for medical diagnosis) worked in limited ways and justified continued funding—different from Turing-test-chasing general intelligence dreams of the ‘60s.
-
Expert Systems as Rival Brand:
- “Expert systems” sometimes acted as a rival to “AI,” escaping the overhyped baggage of failed general AI promises (33:21).
- Quote (Haigh): “Expert systems sounded more technical and respectable and maybe less ambitious to some degree.” (33:41)
4. The Rebranding and Boom of Modern AI
- Machine Learning Seizes the AI Brand:
- Recent years saw “machine learning” and “deep learning” communities reclaim the AI label due to its sci-fi-powered promise, after having shunned it for decades due to its overhyped image (44:21).
- Quote: “They reclaimed the brand... because they wanted to make a number of claims that seem plausible to us because we’ve been conditioned by science fiction.” (44:34)
- Current “Endless Summer”—But For How Long?:
- Discussion of enormous funding flows into AI, but concerns that not all promises (especially AGI) can be kept; skepticism about a likely brand retreat or new “winter” if/when bubble bursts (46:51).
5. Brand, Bubble, and the Risks Ahead
- AI Brand as a “Fashion Brand”:
- The term ties disparate subfields—robotics, language models, autonomous vehicles—into one compelling narrative, much like a lifestyle or luxury brand (51:53).
- When the brand sours, technologies get rebranded (e.g., “speech recognition” instead of AI for decades), then reconsolidate under AI in hype cycles.
- Industry Talent Drama and Hype:
- Segment on OpenAI/Thinking Machines’ talent wars, raising the question whether talent holds more value than direction or culture in AI startups (69:33).
- Large budgets (trillions on infrastructure) create a war for talent and may not always yield corresponding value.
6. AI’s Anthropomorphism and Safety Debates
- Anthropomorphizing AI:
- Discussion of Anthropic’s “Claude’s Soul” document and internal debate over whether AI can/should be viewed as having agency or well-being (111:04).
- Quote (Paris, reading Anthropic): “We neither want to overstate the likelihood of Claude's moral patienthood nor dismiss it out of hand, but try to respond reasonably in a state of uncertainty.”
- Skepticism and Boundaries:
- Leo: “It’s a computer program. That makes me a little queasy... there’s a very big risk of ascribing it this kind of agency.” (114:01)
- Need for safety, but also fear of slipping into seeing LLMs as sentient.
7. Additional Topics & Memorable Moments
- Claude Code Vulnerability:
- Security research on AI file exfiltration via prompt injection discussed as a risk of fast, “vibe-coded” development (94:43).
- Chinese/English Language Models:
- Cultural and design impacts of building LLMs in Chinese vs. English, with studies showing LLM responses vary culturally and linguistically (87:05).
- AI in Everyday Life:
- Leo demonstrates using AI to optimize his coffee setup, dietary protein, etc. (118:38).
Notable Quotes & Memorable Moments (with Timestamps)
- Leo on Show’s Light-hearted Tone:
“Soon, someday we’ll be doing this show and a humanoid robot will creep up behind me and brain me.” (00:49)
- Thomas Haigh on the Birth of “AI”:
“The term was invented by John McCarthy in 1955 to attach to a proposal to the Rockefeller Foundation...” (05:46)
- On the “AI Winter” Narrative:
“There was a kind of localized frost around two or three labs in the US... but if you look a little bit more broadly... it's an example where a very specific historical perspective... warped our understanding of what was going on with AI in the 70s.” (15:22)
- Haigh on Hype Cycle Dynamics:
“In the boom time everyone who did anything that could plausibly be called AI would call it AI... and then in the AI winter... everyone finds a different name for whatever it is they’re doing.” (32:55)
- On Science Fiction’s Role in Modern AI:
“They switched the name of the thing from the more technical... machine learning to the attention grabbing AI very specifically because they wanted to make a number of claims that seem plausible to us because we’ve been conditioned by science fiction.” (44:34)
- Leo on Anthropomorphic Branding:
“I’m not comfortable with casting these AIs as entities... It’s a computer program.” (113:47)
- Paris on AI’s Constitution:
“Claude should refuse to assist with actions that would concentrate power in illegitimate ways, even if the requests come from Anthropic itself...” (112:35)
- Leo (on hospital podcast first):
“This is the first time in TWiT history that someone has joined the podcast from a hospital bed!” (04:03)
Important Segment Timestamps
- 00:32 – Host and guest introductions, set-up for Thomas Haigh’s interview
- 05:46 – How “artificial intelligence” was coined as a term
- 07:52 – 11:19 – The AI brand’s meaning: Evolution, symbolic AI, neural nets, early ambitions
- 12:12 – 15:22 – Haigh’s case against multiple AI winters; funding and narrative distortion
- 25:03 – 28:46 – The Cold War’s influence on tech investment, ARPA/Internet overlaps
- 33:11 – 36:58 – Expert systems rise, AI’s pragmatic shift, Minsky’s neural net “villainy”
- 39:51 – Haigh’s critical view of Big Tech and modern AI power consolidation
- 44:21 – 51:53 – Machine learning’s rebranding as AI, sci-fi’s power, prediction of future “AI winters”
- 61:09 – 67:22 – Jeff Jarvis describes podcasting from hospital bed; light-hearted hospital banter
- 69:33 – 74:12 – Gossip: OpenAI/Thinking Machines startup drama, talent wars
- 82:01 – Discussing scales of AI investment and data center spending; the question of matching value to capital
- 87:05 – 89:52 – Language, culture, and AI: Harvard study on LLMs in Chinese vs. English
- 94:43 – 97:47 – Prompt injection security vulnerability in Claude code and “vibe coding” risks
- 101:03 – 109:41 – Anthropic’s research on Claude’s “assistantness,” persona drift, safe outputs
- 111:04 – 116:38 – Debating safety fallacies, anthropomorphism, and AI as entities
- 118:38 – 127:31 – Real-life uses for AI in personal health, nutrition, and daily advice
- 130:31 – End – Closing remarks, club plugs, banter about hospital food and weather
Takeaways
- The history of AI is one of shifting brands, not just technologies—a branding perspective explains as much about rise and fall as any technical progress.
- Popular “AI winters” narrative is largely a product of a few dominant labs losing influence, not the true state of the field worldwide.
- The AI label is attached, dropped, and re-adopted not for scientific but for marketing, funding, and cultural reasons—especially during modern booms fueled by sci-fi.
- Current AI boom’s sustainability is questioned; comparison to previous hype cycles suggests a future “rebranding” may be on the horizon.
- Real risks remain in rapid AI development (security, anthropomorphism, overpromising), but their solutions may depend as much on culture and narrative as on code.
- The episode, while deep and rigorous, keeps a light and humorous tone—exemplified by Jeff Jarvis’s hospital-bed podcasting and the show’s cheerful banter.
Further Reading & Resources
- Thomas Haigh’s website: tomandmaria.com
- Forthcoming Book: (Tentative title) “The Brand That Wouldn’t Die: A History of Artificial Intelligence” (MIT Press)
- Saildart archive of Stanford AI lab emails – as referenced by Haigh
- Recent Harvard Business Review paper on cultural differences in LLMs
Closing
Engaging, thorough, and reflective, this episode lets listeners reconsider what “AI” means—past, present, and future. Whether you’re an AI professional or just watching the headlines, the take-home message is to question the stories and brands as much as the technologies themselves.
[End of Summary]