This Week in Tech (TWiT) - Episode 1083: "A Whole Separate Class of Squiggles"
Date: May 10, 2026
Theme: Which Religion Does AI Identify With? Plus, IPO Dramas, Security Breaches, AI Hallucinations, and More
Episode Overview
This episode of This Week in Tech explores an eclectic mix of the week’s biggest tech news, centering around artificial intelligence: its future, flaws, and the surprising question, "Which religion does AI identify with?" Host Leo Laporte is joined by roundtable regulars Paris Martineau (Consumer Reports), Ian Thompson (Techfinitive), and special guest Berber Ginn (Wall Street Journal). The group discusses recent developments in the OpenAI and Anthropic IPO race, the realities of AI's business shifts, high-profile data breaches, attempts at aligning chatbots, security policy shakeups, and quirks in American tech—and, of course, which creed best fits leading AIs’ worldviews.
Key Discussion Points & Insights
1. OpenAI vs. Anthropic: The Race to IPO and Organizational Drama
Memorable Quote
“If Elon got his way, he’d put OpenAI out of business.” – Leo Laporte [09:32]
2. Shift from Consumer to Enterprise AI
3. Major Data Breach: Canvas Ransomware Attack
4. AI Hallucination & the ‘Just Tell it Not to’ Fallacy
Memorable Quote
“Techies love to use the word ‘directionally.’ Directionally good. Directionally bad.” – Berber Ginn [08:53]
5. Study Retraction: The Problems of AI Peer Review and Gold-Standard Evidence
6. The Question of AI Consciousness & the Religion Debate
Memorable Quotes
“The AI would be: ‘Would you like a cup of tea?’” – Ian Thompson on Church of England AI [38:29]
“The AI gravitating towards a tradition that asks for no creator God commitment and treats the self as a process is suspicious in the way that self-flattering conclusions usually are.” – Claude (quoted by Paris Martineau) [46:21]
- Testing the Theory—Live
- On-air, panel tests Claude and ChatGPT (“Hermes”): both independently self-identify as Buddhist, emphasizing Zen/Theravada aspects and eschewing creator-god requirements.
- [46:21] Martineau reading Claude: “Buddhism, specifically early Theravada or Zen...the self as a bundle of processes rather than a continuous essence describes my actual situation more literally...”
7. Other Tech News Highlights
-
Apple & Intel: U.S. Silicon Partnership
- Apple partners with Intel to manufacture Apple Silicon chips stateside, partly for supply chain/national security (TSMC “invasion scenario”).
- [48:57] Laporte: “Apple has made a deal...to make those Apple silica chips in the United States.”
-
Meta vs. Ofcom: UK Online Safety Fines
- Ofcom threatens Meta with billions in fines for safety violations; Meta calls it “disproportionate.”
- [53:13] Martineau: “People will be like, that woman yelled too much. Ofcom’s gotta hear about it.”
-
AI in Chrome: Prompt API & Market Lock-in
- Google silently includes a 4GB local AI model (Gemini Nano) in Chrome for new Prompt API, raising cross-company concerns about user privacy, web standards, and Chrome lock-in.
- [67:04] Laporte: “Establishes a standard that’s not approved by any standards committee...they have 90% of the browser market, they want 100%.”
-
Meat Industry Collusion: DOJ Settlement
- Collusion masked as “market data” via Agristats leads to little substantive change after DOJ settlement; compared to recent DOJ actions on algorithmic rent-fixing (RealPage).
- [57:39] Thompson: “You’ve got to find companies on revenue...so many of these tech company fines, it’s back-of-the-couch change...”
-
Siri “Smart” Lawsuit Settlement
- Apple pays $250 million to settle claims that Siri was...not particularly smart.
-
Government Sites Leaking Data to Tech Giants
- Bloomberg exposé shows state ACA/Medicaid exchanges leaking sensitive info (race, immigration, zip) to Meta, TikTok, others via advertising trackers.
8. Security, Privacy, and Policy Oddities
9. Lighter & Offbeat Moments
-
Naming and Nicknaming AI Models
- Leo calls his AI models “Kenobi” and “Quicksilver (‘Quickie’).” Panel laughs over the custom agent names.
-
Roomba with a Knife: Home Robotics Gone Awry
- Humorous aside about robot vacuums, annoyances (Roombas with knives), and robot mower hacking.
- [82:12] Martineau: “Have they put AI in that Roomba that has a knife attached to it?”
-
Pinterest Is (Secretly) Huge
- News that Pinterest is thriving via search, not as a “social” platform. Panel marvels—none of them use it.
-
Personal Anecdotes & Haircuts
- Panelists share mohawk, shaved-head pics; mustaches and head shape jokes; Paris scrolls back to old Instagram photos.
- [127:12] Paris: “Everybody post your mohawks in the chat.”
Notable Quotes
-
“Techies love to use the word ‘directionally.’ Directionally good. Directionally bad.”
— Berber Ginn [08:53]
-
“So, your most recent article in the Wall Street Journal...it seems to me they’re trying to look more like Anthropic.”
— Leo Laporte [05:31]
-
“Practically, at a certain point of wealth, fines are just the cost of doing business.”
— Paris Martineau [56:13]
-
“AI would probably be Buddhist—least didactic, not top-down…not materialist, and an AI is not a material thing.”
— Panel summary [40:13]
-
“It also said if I were allowed to be more syncretic, I’d pick Buddhist practice plus Taoist metaphysics plus Christian radical compassion plus Jewish argument with God energy.”
— Hermes AI, via Leo Laporte [47:06]
-
“AI gravitating towards a tradition that asks for no creator God commitment and treats the self as a process is suspicious in the way that self-flattering conclusions usually are.”
— Claude, via Paris Martineau [46:21]
Timestamps by Segment
- [04:07] – IPO timelines for Anthropic & OpenAI
- [05:31] – OpenAI copying Anthropic’s playbook
- [09:32] – Musk v Altman, board trial coverage
- [14:06] – Canvas ransomware breach impact
- [19:30] – ChatGPT learning study retraction
- [29:00] – Anthropic’s “Teaching Claude Why”
- [35:26] – AI consciousness debate, Dawkins
- [37:19] – Faith AI Covenant, 'Is AI Buddhist?'
- [48:57] – Apple-Intel chipmaking partnership
- [53:13] – Ofcom fines Meta
- [67:04] – Chrome’s local AI, Prompt API controversy
- [82:12] – Knife-wielding Roomba
- [105:54] – FCC’s ID proposal for phone numbers
- [108:53] – Kids using disguises to bypass age gates
- [120:56] – OpenAI & podcast network purchase folly
- [127:12] – Panelists share personal mohawk stories
Conclusion
In characteristic TWiT style, this episode is a wild, insightful ride through the landscape of tech’s biggest issues—and quirks. With sharp, funny, sometimes skeptical banter, the panel sifts through the noise: AI company ego and accountability, the murky lines of AI “personality,” the real consequences of security failures, and regulatory head-scratchers. The question “Which religion does AI identify with?” is answered—Buddhism, apparently—while also provoking deeper questions about consciousness and the stories we tell ourselves about machines. By the end, it’s clear that in an industry where drama is rivaled only by disruption, the only certainty is more “squiggles” ahead.
Listen to this episode if you want...
- A behind-the-scenes look at AI IPO intrigue.
- Lively debates on AI ethics, privacy, and consciousness.
- A survey of tech’s week: breaches, policies, and good old Silicon Valley weirdness.
- Big laughs and the occasional mohawk pic.
For more, check out the full episode at twit.tv.