Research: The AI-Worldview Poles -- Different Pawls, Same Ratchet
Source dossier for The Ratchet (Book 2) -- Ch07 'The Priest' and Ch20 'The
Contents
Source dossier for The Ratchet (Book 2) – Ch07 “The Priest” and Ch20 “The Blueprint” – and seed material for the Book 3 “creeds” thread. The thesis the draft below argues: the loud public fight is a false binary (accelerate vs. extinction) that crowds out the present-harm and dignity voices, and the crowding-out is itself a control-grid artifact, because it keeps the argument fixed on the future while the apparatus gets built in the present. The camps are not opponents in any way that matters to the machinery; they are competing justifications that all turn the same wheel. Different pawls, same ratchet. The Vatican’s dignity-of-the-human-person frame is presented as the documented third pole – the alternative creed, not endorsed – because it is the one that names what the binary erases: the irreducible human.
PERISHABILITY NOTE: every 2025-2026 claim below is dated and flagged (Magnifica Humanitas verified live + verbatim on vatican.va, 2026-07-14). The single most perishable item is the Magnifica Humanitas encyclical (15 May 2026) – it is weeks old as of this writing. RE-VERIFY AT PRINT.
VERIFIED SOURCES
Accelerationist / e/acc
- Marc Andreessen, “The Techno-Optimist Manifesto,” a16z, posted 16 October
2023. Primary: a16z
- Confirmed content: ~5,200 words; the phrase “We believe” repeated 113 times; calls AI a “philosopher’s stone”; frames technology (not population or resources) as the only perpetual source of growth; explicitly names “effective accelerationism” / e/acc lineage.
- Wikipedia overview (corroborating, dated): Wikipedia
- Contemporaneous coverage: Fortune
- e/acc as a movement (Beff Jezos / Guillaume Verdon) is named in the manifesto’s lineage. The Verdon unmasking and the broader e/acc subculture are well documented but the specific attributions are best cited to the manifesto itself, which is stable. VERIFY any Verdon-specific quote separately before print – those are forum/Twitter-sourced and perishable.
Doomer / x-risk
- Center for AI Safety (CAIS), “Statement on AI Risk,” 30 May 2023. One
sentence: “Mitigating the risk of extinction from AI should be a global
priority alongside other societal-scale risks such as pandemics and nuclear
war.” Primary: Center for AI Safety
- Signatories include Geoffrey Hinton, Yoshua Bengio, Sam Altman (OpenAI), Demis Hassabis (DeepMind), Dario Amodei (Anthropic), plus Microsoft and Google executives. Notably NOT Meta/LeCun. Wikipedia (corroborating): Wikipedia
- Coverage: TechCrunch
- Geoffrey Hinton extinction probability estimate: stated “10 to 20 percent”
chance AI leads to human extinction within ~30 years, on BBC Radio 4 Today,
late December 2024 – a revision UP from his earlier ~10 percent figure.
- Coverage (dated): Tech Times
- The Guardian carried the same “10-20%” framing.
- FLAG: the “10-20%” number is an off-the-cuff interview estimate, not a paper. Cite it as what Hinton SAID, with the date and outlet, never as a measured quantity. RE-VERIFY AT PRINT – he revises this number in public repeatedly and it drifts.
- Hinton resigned from Google May 2023 specifically to speak freely about these risks (cross-references the CAIS statement timing).
Vatican / dignity (the third pole)
- “Rome Call for AI Ethics” (formally “Call for an AI Ethics”), signed in Rome 28 February 2020. First signatories: Pontifical Academy for Life (Abp. Vincenzo Paglia), Microsoft (Brad Smith), IBM (John Kelly III), FAO (Qu Dongyu), and the Italian Ministry for Technological Innovation (Paola Pisano). Six principles: transparency, inclusion, responsibility, impartiality, reliability, security/privacy. Primary: Rome Call
- “Antiqua et Nova” – Note on the Relationship Between Artificial Intelligence
and Human Intelligence, joint doctrinal note of the Dicastery for the
Doctrine of the Faith and the Dicastery for Culture and Education, dated
28 January 2025. 117 paragraphs. Primary (vatican.va):
Vatican
- Confirmed content: draws a hard line between what machines do and what the human mind IS (intelligence as moral/spiritual openness to truth – conscience, responsibility, soul); flags autonomous lethal weapons as grave ethical concern citing Francis on “existential risk”; warns AI in education “merely provide[s] answers instead of prompting students to arrive at answers themselves” (this line is a direct Book-3 “Mind / cognitive capture” bridge – the Church named cognitive offload as a formation risk in a doctrinal note).
- Vatican News: Vatican News
- Pope Leo XIV (Robert Francis Prevost, elected May 2025). Name rationale:
stated to the College of Cardinals two days after election that he chose the
name partly to honor Leo XIII, whose “Rerum Novarum” (1891) addressed “the
social question in the context of the first great industrial revolution,”
and that AI poses “new challenges for the defence of human dignity, justice
and labour.”
- Vatican News on the cardinals address (10 May 2025): Vatican News
- National Catholic Reporter (frame: church-defended-workers-from-industry, now Leo takes on AI): National Catholic Reporter
- Magnifica Humanitas – Pope Leo XIV’s FIRST encyclical, “On Safeguarding
the Human Person in the Time of Artificial Intelligence.” SIGNED 15 May 2026
(the 135th anniversary of Rerum Novarum); RELEASED 25 May 2026. Primary
(vatican.va, confirmed live and read):
Vatican
- Confirmed opening line: “Humanity, created by God in all its grandeur, is today facing a pivotal choice” – framed as Tower of Babel vs. the city in which God and humanity dwell together.
- Confirmed line: “technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate, and use it” – this is a near-verbatim statement of The Ratchet’s own thesis from inside the Vatican, and is the single best pull-quote for Ch07/Ch20.
- Coverage: TIME TIME ; Washington Post Washington Post ; full text mirror at National Catholic Register National Catholic Register
- CRITICAL CORRECTION + PERISHABLE FLAG: the brief/spine assumed Leo’s AI framing was at the “statements” stage and dated the encyclical 2025. It is a signed 2026 encyclical (15 May 2026). This is the freshest claim in the dossier – weeks old. RE-VERIFY AT PRINT, and confirm the encyclical title/date have not been superseded by an official Latin editio typica with a different paragraph numbering before quoting any paragraph number.
AI-ethics / present-harms (x-risk-is-hype-laundering camp)
- Bender, Gebru, McMillan-Major, “Shmargaret Shmitchell” (Margaret Mitchell),
“On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?,”
ACM FAccT 2021. PDF: FAccT 2021 (PDF)
- Wikipedia overview: Wikipedia
- Gebru and Mitchell were fired from Google in the fallout (Gebru late 2020, Mitchell early 2021). This is the camp’s origin wound and is well sourced.
- The camp’s argument (Gebru, frequently restated): framing AI as an
autonomous “superintelligence” lets the humans who build, fund, and profit
from it off the hook, and the hype – utopian OR apocalyptic – is a
marketing engine that distracts from present harms (dataset bias,
exploited data labelers, copyright, environmental cost). This is the
load-bearing “x-risk launders present harm” claim. Cite to Gebru’s own
public statements; the cleanest is the DAIR Institute framing and her
“Tech Won’t Save Us” interview: Tech Won’t Save Us
- FLAG: attribute the “hype laundering” framing to Gebru/DAIR as their POSITION, not as our editorial voice (per the no-editorial-framing rule).
- Joy Buolamwini + Timnit Gebru, “Gender Shades,” 2018. Commercial facial
analysis error rate up to 34.7% for darker-skinned women vs. 0.8% for
lighter-skinned men. MIT News: MIT News
- Project page: MIT Media Lab
- Buolamwini founded the Algorithmic Justice League. Wikipedia
- Function: the concrete, measured present-harm receipt that the x-risk debate talks over.
National-security realists (race-with-China)
- National Security Commission on Artificial Intelligence (NSCAI), Final
Report, March 2021. Chair: Eric Schmidt; Vice Chair: Robert Work. Core
claim: the U.S. is “not prepared to defend or compete” against China and
could lose AI leadership within a decade. Two parts: “Defending America in
the AI Era” / “Winning the Technology Competition.”
- Primary (GPO GovInfo, stable mirror – the nscai.gov domains went dark when the commission dissolved): GovInfo
- Original report portal (site now offline; Wayback snapshot): Wayback
- Wikipedia: Wikipedia
- Note for the draft: the report itself cites China’s AI surveillance of Uyghurs as the cautionary precedent – i.e., the realist camp’s own document concedes the apparatus is the destination; it argues only over WHOSE apparatus. Useful “same ratchet” receipt.
- Adversarial counter-source (Schmidt conflict-of-interest critique), keep for balance: Tech Transparency Project
Refusers
- Meredith Whittaker (President, Signal). Position: agentic AI is an
“existential threat” to secure messaging and to anyone building apps,
because agents demand deep access (bank details, passwords) and create a
massive new attack surface; OS-level agent integration is “very reckless.”
Coined “putting your brain in a jar.”
- TechCrunch (SXSW, 7 March 2025): TechCrunch
- Fortune (27 Nov 2025): Fortune
- “Brain in a jar” is the perfect bridge to Book 3’s cognition-capture thesis – the refuser names cognitive externalization as the threat, from the security side.
Pragmatist governance
- EU AI Act – Regulation (EU) 2024/1689. Entered into force 1 August 2024.
Four risk tiers: unacceptable / high / limited / minimal. Fines up to EUR 35
million or 7% of global annual turnover. High-risk obligations (biometrics,
critical infrastructure, employment, migration) phase in through 2 December
2027.
- Commission “enters into force” notice: European Commission
- Regulatory framework portal: European Commission
- FLAG: the phase-in dates are perishable and have already slipped once in EU practice. RE-VERIFY AT PRINT.
- NIST AI Risk Management Framework (AI RMF 1.0), NIST AI 100-1, released 26 January 2023. VOLUNTARY / non-binding. Four functions: GOVERN, MAP, MEASURE, MANAGE.
DRAFT SECTION – “Different Pawls, Same Ratchet”
(The Ratchet voice: third-person institutional, sardonic but fair, all sides present, receipts in prose. ~1,050 words. Drop-in candidate for Ch07 “The Priest” with the closing turn seeding Ch20 “The Blueprint.”)
There is a debate about artificial intelligence, and you have heard it. It has two sides. On one side stand the accelerationists, who believe the machine is a philosopher’s stone and that the only sin is slowing down. Their scripture is a five-thousand-word document Marc Andreessen posted to his venture firm’s website in October 2023, in which the words “We believe” appear one hundred and thirteen times, technology is named the only perpetual source of human growth, and the people who counsel caution are described, more or less, as enemies of life. On the other side stand the doomers, who believe the machine may kill everyone. Their scripture is a single sentence, published by the Center for AI Safety in May 2023 and signed by Geoffrey Hinton, Yoshua Bengio, and the chief executives of every major laboratory building the thing: that mitigating “the risk of extinction from AI” belongs on the same shelf as pandemics and nuclear war. Hinton, who quit Google the same month so he could say it out loud, now puts the odds of human extinction at ten to twenty percent and revises the figure upward when asked.
It is a thrilling argument. Heaven on one side, the apocalypse on the other, and a clean line down the middle for the cameras. It is also, considered structurally, the most useful argument the apparatus has ever been handed, because both sides are arguing about the future.
While the future is litigated, the present is built. And the present has its own witnesses, who are mostly not invited to the debate. In 2018 Joy Buolamwini and Timnit Gebru measured commercial facial-recognition systems and found they misread darker-skinned women up to thirty-five percent of the time and lighter-skinned men eight-tenths of one percent of the time – not a hypothetical superintelligence, a shipping product, wrong about real faces in a measurable direction. In 2021 Gebru and Margaret Mitchell warned, in a paper on the dangers of large language models, that the systems were unaccountable and biased by construction; Google fired them both. Gebru’s argument since has been blunt and inconvenient to everyone: framing the machine as an autonomous god-in-waiting, whether savior or destroyer, is precisely what lets the humans who design it, fund it, and bill for it walk away clean. The hype, she says, is the marketing – and apocalypse sells as well as utopia. This is the present-harm camp, and its central complaint is not that the extinction debate is wrong. It is that the extinction debate is load-bearing: it keeps the conversation pointed at a horizon while the wiring goes in under the floor.
There are other voices, each with a real position and each easy to lose in the noise. The national-security realists – Eric Schmidt’s commission chief among them – argue that the only question that matters is whether the surveillance apparatus ends up American or Chinese; their own 2021 report cites the mass-monitoring of the Uyghurs as the thing to avoid, which is to say it concedes the destination and argues only over the management. The refusers, led by Signal’s Meredith Whittaker, decline the premise entirely: an AI agent, she says, is a brain in a jar, and handing it your passwords and your bank login to save yourself some typing is a security catastrophe dressed as convenience. And the pragmatists – the EU with its risk tiers and seven-percent fines, NIST with its voluntary four-function framework – produce the paperwork, the grammar of “trustworthy AI” and “acceptable risk” that the apparatus then adopts as its own compliance vocabulary, the way a body learns to issue its own permits.
What none of them quite says, except one, is what the thing being lost actually is. That pole is occupied, improbably, by the Catholic Church – which is presented here not because it is right but because it is the only institution that has put on paper, in plain language, the stake the binary erases. The Rome Call for AI Ethics, signed in 2020 by the Pontifical Academy for Life alongside Microsoft and IBM – the same corporations that build the apparatus, a detail worth holding – set out six principles. The 2025 doctrinal note Antiqua et Nova drew the line the engineers will not: that intelligence is not what a machine does but a moral and spiritual openness to truth, and warned, in a Vatican document, that an education which “merely provide[s] answers instead of prompting students to arrive at answers themselves” deforms the people it claims to serve. Then in May 2026 a new pope who had taken the name Leo – explicitly to invoke Leo XIII, who met the first industrial revolution with a defense of the worker – signed his first encyclical, Magnifica Humanitas, on the 135th anniversary of that defense. Its thesis fits in one of its own lines: “technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate, and use it.” It is, word for nearly identical word, the argument of this book, arriving from the one institution old enough to have watched a machine eat a way of life before.
The Church’s frame is the dignity of the human person, and the irreducible human – the thing that is not a confidence score, not an output, not a risk tier. Take it or leave it as theology. Note only this: it is the single pole in the entire argument that is not about whether the machine arrives faster or slower, saves us or kills us. It is about who is in the room when it does.
That is the tell. The accelerationist wants the ratchet turned faster; the doomer wants a brake on the same ratchet; the realist wants his hand on the crank instead of Beijing’s; the pragmatist wants the crank certified to standard. Four positions, four justifications, one wheel, and a wheel that – by design – only turns one way. Different pawls, same ratchet. The blueprint that follows does not take a side in the debate. It points at the machine the debate is standing on.
Related research
- The AI Governance Ratchet — the instruments these poles fight to operate
- AI-worldview camps — the fuller 7-camp map this frames
- AI governance tracker · Quiet Autocomplete