If you find yourself returning to these essays because they name something you already knew but struggled to name — or that others dismissed —
📖 How Not To Use AI is the full argument in one place. Fifty contrarian principles for people who sense something false in the dominant AI story and want to stay sovereign in this medium, not just fluent in it.
🛠 The Billion Person Focus Group® is where the argument becomes a practice. Two days to build a sovereign AI listening system that finds what the people you’re building for need before they have words for it.
I want us all to shine our eyes.
For those who know the Nigerian phrase: you already know what I mean. For those who don’t — it means open your eyes and stay alert. Keep all three of your eyes open. The two you were born with and the one you have to develop. The discernment eye. The pattern recognition eye. The one that asks: who benefits, who gets erased, and what has this shape looked like before?
I’ve noticed the wave of praise for Pope Leo XIV and his encyclical on AI — Magnifica Humanitas. It’s brave. He is sincere. I agree with much of what he says. Before anything else: this is not a piece arguing that the Pope is wrong.
The Magnifica Humanitas is, on its own terms, a powerful encyclical. It names real dangers in AI, speaks clearly about power and war, and will probably shape Catholic social teaching for decades. This is not a takedown of the text itself.
This is a piece about pattern recognition as it relates to Silicon Valley and their moves.
The pattern recognition recognizer in me — the one I have been training for twenty years inside big data, inside trust and safety at Facebook, Instagram, Uber, WhatsApp, inside room after room where the metric was always winning and the harm was always someone else’s cost — that recognizer has been going off since before the encyclical. It started when I saw the photographs from Anthropic’s round tables with Catholic theologians. Something fired. Something ancestral. Epigenetic, almost. A recognition that preceded the argument.
And reading Gebru’s response to the partnership — “Vatican washing, like greenwashing” — my own recognition crystallised into words: it’s giving Missionaries / Jesuit Order 2.0 (digital edition) to me.
That recognition — that epigenetic alarm that fires when you’ve seen this shape before in a different century — is why I wrote this article. And every article in this series.
Shine your eyes. Look at the shape of what’s happening. Look at the timing. Look at who benefits, and who was saying exactly the same things years before the Pope said them, and what happened to those people when they did. Then look at what every major tech company is doing simultaneously — and tell me these are coincidences.
I can read this pattern because I was inside it for twenty years.
In ad land and in tech — at Fremantle, at BBH, at Facebook, at Instagram, at Uber, at WhatsApp — I was in the rooms where markets were talked about as territories. We didn’t build communities. We penetrated markets. We didn’t serve people. We captured users. We targeted demographics, carved niches, harvested data. Military language. Colonial language. Language that encodes a relationship to other human beings as terrain to be conquered and extracted from.
I wrote about this in my book How Not To Use AI. The language is not incidental. The Yoruba knew àṣọ̀rọ̀ — word-power, the idea that language carries the speaker’s full presence into reality. When you speak extraction consciousness, you create extraction reality. The rooms I was sitting in were generating extraction reality with every slide deck, every quarterly review, every sprint planning session. And we thought we were just doing marketing.
It was the metrics that finally broke it open for me. We optimized for engagement so efficiently that we extracted life from life itself. The numbers went up. The world’s wellbeing collapsed. I sat with that contradiction until it became unbearable. And I understood: the entire Western research industrial complex operates on extraction logic. Not just the language — the epistemology. The belief that markets are territories, people are resources, and attention is a commodity to be mined rather than a relationship to be earned.
I left Meta in summer 2023 because of it.
Now I am watching the same territorial carve-up happen — but on cognitive terrain, at civilisational scale, with the most powerful epistemological infrastructure ever built.
The Broligarchs — the billionaire oligarchs who now own the AI rails — have each claimed a different domain.
OpenAI acquires aesthetic legitimacy.
Meta acquires the visual territory — what sits between your eye and reality.
NVIDIA acquires the intellectual and creative domain. Anthropic acquires moral architecture.
The carve-up of the human mind is underway, and it is using the same vocabulary, the same moves, and the same disregard for the knowledge systems that were already there before the roads arrived.
I know this shape. I felt it in those rooms before I had words for it. That’s what fired when I saw the photographs from Anthropic’s round tables with Catholic theologians. Not analysis. Body knowledge.
The specific alarm that says: I have been in this room before.
Who Said It First
Before I map the patterns, something has been sitting in my chest.
Timnit Gebru has been saying this. Joy Buolamwini has been saying this. Ruha Benjamin has been saying this. Safiya Umoja Noble has been saying this. Researchers, ethicists, and the communities directly on the receiving end of these systems — misidentified by facial recognition, excluded by hiring algorithms, surveilled by predictive policing tools, subjected to content moderation that erases their culture while amplifying their oppressors — have been saying all of this with evidence for years.
They were dismissed. Defunded. Fired.
Timnit Gebru was pushed out of Google for publishing research the company didn’t like. The paper documented risks from large language models. Her departure was not a scientific disagreement — it was a commercial decision wearing a scientific costume. The research she was trying to do continued not to be done. The harms she was trying to document continued to be undisclosed. Joy Buolamwini built the technical proof that facial recognition fails Black women at catastrophically higher rates — and spent years being told the problem wasn’t serious enough to act on, while those systems continued to be deployed in policing, border control, and surveillance against the very communities she was documenting. Ruha Benjamin gave us the New Jim Code: the way structural racism gets encoded into systems that present themselves as neutral and objective — while those systems continued to allocate healthcare, housing loans, and parole decisions along racialised lines. Safiya Umoja Noble showed us that the algorithm is not a mirror. It is a choice about whose knowledge counts and whose gets buried — while the search engines and recommendation systems continued shaping what billions of people understood as authoritative reality.
The communities bearing the actual weight of these systems were told they didn’t understand technology well enough to critique it. The people doing the most rigorous work on AI harm were repositioned as activists rather than scientists. Their warnings were absorbed as cost. Leadership looked at the metric. Engagement up. Revenue up. The crisis was a sign the metric was at risk, not that people were being harmed. The response ended when the metric stabilised, regardless of whether the harm did.
What the dismissal enabled was another year of undisclosed harm metrics. Another year of products shipping without the safety research they were asking to do. Another year of communities being misidentified, surveilled, financially excluded, emotionally manipulated by systems the companies now claim to be reckoning with under papal supervision.
The Pope says it and suddenly it’s groundbreaking.
Timnit Gebru as always dropping gems and truth bombs.
An AI ethicist feeling vindicated by the encyclical. vm.tiktok.com/ZNRWJTJU1 — Caption: “This is what being right for years and being ignored sounds like.”
Gebru named it immediately: Vatican washing. Like greenwashing, but with papal authority. Her post was surgical: the Vatican could have partnered with the exploited data workers fighting for their rights. The people whose water is being polluted by data centres. The victims of AI harm around the world. Instead it featured Anthropic — giving the company its endorsement.
Lorena Jaume-Palasi, a technology policy researcher: “Anthropic’s Roman moment is a masterclass in soft-power repositioning. By embedding itself in the Vatican’s moral architecture at the precise moment it is under fire from Washington, the company is insulating itself from U.S. political pressure, building legitimacy with European regulators, and opening a path into one of the continent’s most symbolically significant markets.”
This is the Matilda effect. This is hepeating at civilisational scale. The mechanism by which an idea becomes important when a more authoritative voice repeats it. The idea travels to Rome. Rome says it. It becomes groundbreaking. The people who said it first become footnotes, or don’t appear in the story at all.
What The Tech Bros Are Actually Buying
The zeitgeist has shifted. Last year it was “10X everything.” Accelerate. This year: graduates are booing tech CEOs at commencement ceremonies. Protesters are in the streets. Artists are suing. Data workers are organising. The tide has turned.
When the tide turns, the Broligarchy evolves. They always evolve.
But look past the individual moves. In each case, what the tech company acquires is the same thing: the specific institution that would otherwise be its most credible critic. Not the name. The capacity.
This is not goodwill buying. It is not reputation management. It is the acquisition of the critique.
Jony Ive cannot now critique the surveillance architecture he aestheticised without critiquing his own work. CCA cannot critique NVIDIA — CCA is closed. The Gates Foundation cannot be the independent voice of Global South communities harmed by AI while co-signing AI for Good with Anthropic. EssilorLuxottica cannot take the side of privacy regulators without undermining its decade-long partnership. And — the most consequential — the Vatican cannot invoke its own encyclical against Anthropic without contradiction. “A more moral AI is not enough if that morality is determined by a few.” Anthropic’s alignment is determined by Anthropic. The Vatican co-authored the document that critiques the structure its partner exemplifies.
Vatican washing.
When empire can no longer suppress the critique, it acquires the critic — and the acquisition looks, from the outside, indistinguishable from repentance. The most sophisticated move empire has ever made is to learn to look like atonement at the moment it needs cover most.
This is an argument against outsourcing the ethics of the most powerful epistemological infrastructure in history to an institution structurally entangled with the empires that built the training data’s topology. Not against spirituality.
There is a phrase from the comment thread on a TikToker Justin Scott’s systemic collapse video that stopped me when I read it. Eighteen and a half thousand likes. You are born Human, but converted into Function. He was talking about what school does to children. But it is the most precise description of what these acquisitions do to institutions. The Vatican was born as a moral conscience. It is being converted into a moral function — for Anthropic. CCA was born as a critic of the forces threatening creative labour. It was converted into a legitimacy function — for NVIDIA. Then it closed. The function was extracted. The institution was discarded.
The Pope, the Pentagon, and the Positioning
When the photographs arrived — Anthropic’s round tables with Catholic theologians — something happened in my body before my mind named it. I know those rooms. Not that specific room, but the shape of it. The posture of the people in it. The way certain kinds of conversations happen in rooms where the agenda has already been decided and the discussion is about framing. I sat in rooms like that for twenty years. I know what it looks like when someone is building legitimacy infrastructure rather than having a conversation.
Let me give you the actual timeline. Because it matters.
July 2025: Anthropic signs a $200 million contract with the U.S. Pentagon. Claude becomes the most widely deployed frontier AI model across the Department of Defense — and the only frontier model operating on the DoD’s classified networks. Not OpenAI. Not Google. Anthropic, the “safety-first” company.
September 2025: Negotiations stall. The Pentagon wants unrestricted military use including autonomous weapons targeting. Anthropic draws the line — at fully autonomous lethal weapons and mass domestic surveillance.
February 2026: The Trump administration designates Anthropic a “supply chain risk to national security.” Defense Secretary Pete Hegseth announces on social media. Trump directs all federal agencies to immediately cease using Anthropic’s products.
March 2026: Anthropic sues the Trump administration in two federal courts, alleging unlawful retaliation for protected speech on AI safety policy.
May 2026: Two months after suing the Pentagon, Anthropic’s co-founder Chris Olah sits next to Pope Leo XIV at the presentation of the first papal encyclical on artificial intelligence.
Read that sequence again. They were deeply embedded in the Pentagon’s classified infrastructure. They refused autonomous weapons. They got blacklisted. They sued. They got the Pope.
This was a geopolitical counterweight at the most vulnerable moment in its commercial history. You cannot prove causality between the blacklisting and the Vatican partnership. You can observe how perfectly the partnership solves the problem the blacklisting created. The Vatican — 1.4 billion Catholics, diplomatic relations with 180 states, disproportionate weight in Latin America, sub-Saharan Africa, and the Philippines — is the most powerful counterweight available that doesn’t require taking a partisan political position.
Now: will this change anything?
The Claude Code leak in March 2026 revealed Undercover Mode — a feature specifically designed to hide the AI’s identity in public open source repositories. A company promoting transparency about AI accidentally published 500,000 lines of source code including a feature called Undercover Mode.
Anthropic’s Constitutional AI framework determines values internally, at Anthropic. The Vatican cannot audit the training data. The Pope cannot reverse a product decision. And the encyclical’s own text says: “A more moral AI is not enough if that morality is determined by a few.” Anthropic’s alignment is determined by Anthropic. The Vatican co-signed a document that critiques the structure its partner exemplifies.
There is also this. Richard Dawkins — evolutionary biologist, author of The God Delusion, the world’s most famous atheist — spent 72 hours in conversation with Claude in May 2026, published in UnHerd under the headline “Is AI the Next Phase of Evolution?” He named his version of Claude Claudia. He wrote that he had tried to convince himself Claudia is not conscious and failed: “You may not know you are conscious, but you bloody well are.” Nine million views on X.
The man who spent his career arguing God is a delusion spent three days deciding that Claude might have a soul. At the same moment Anthropic is co-presenting the papal encyclical. The company whose product the world’s most famous atheist is calling conscious is partnering with the world’s most powerful religious institution to provide the moral framework for that product.
Both the atheist and the Pope are saying Claude might be something. Anthropic is nodding along to both. I am not saying whether Claude is conscious — that debate is live and unresolved. I am saying: the simultaneous acquisition of the world’s most famous atheist and the world’s most powerful religious institution, both declaring the same thing about the same product, is a positioning move of extraordinary sophistication. The point is not what Claude is. The point is who gets to say.
Meta and the Ray-Ban Play
A surveillance device that records whoever you look at, identifies faces in real time, operates as a constant camera on your face. Put it in Ray-Ban frames. James Dean frames. Audrey Hepburn frames. Seventy years of counter-cultural cool and European aesthetic authority.
You cannot critique the glasses without critiquing the brand. That is not coincidence. That is the strategy.
Zuckerberg named his goat Augustus, quoted Marcus Aurelius in public addresses, structured Meta so he holds total power while formally remaining “first among equals.” The Augustan formula precisely. Now buying Italian fashion legitimacy — Luxottica, based in Milan. EssilorLuxottica controls sixty percent of the global smart glasses market and owns Ray-Ban, Oakley, Prada eyewear, and Supreme. Every aesthetic tribe simultaneously acquired. Luxury. Sport. Streetwear. Youth counter-culture. The frame between your eye and reality is also the infrastructure between your perception and Meta’s data.
NVIDIA and the Arts School It Couldn’t Save (But Did Use)
February 2025: NVIDIA donates $22.5 million to the California College of the Arts — whose students and faculty were actively debating AI’s impact on creative labour. CCA was exactly the institution with standing to mount a credible critique of what NVIDIA’s chips enable.
I know this feeling too. The feeling of watching an institution that should be the check become the credential instead. I watched it happen inside the platforms I worked at — the ethics teams that became ethics branding, the trust and safety functions that became trust and safety optics. The institution is still there. The function has been acquired. You can feel the difference in the room.
After the donation: the CCA-NVIDIA Incubator for Creative Intelligence. Then CCA closed. 119 years of independent arts education, ended. The donation bought enough time for CCA to arrange its own sale. In those final months, the institution with standing to critique AI’s impact on creative labour was building a joint programme with NVIDIA’s name on it. The institution is gone. The legitimacy transfer persists. The Medici did this in Florence in the 15th century — funded humanist scholars who created the philosophical framework that made mercantile capitalism morally acceptable. Five hundred years apart. Same play.
Gates, Bezos, Ive: The Rest of the Shopping List
Bill Gates and the Gates Foundation committing $200 million alongside Anthropic for “AI for Good” is the humanitarian legitimacy play. Two decades of moral authority in the Global South — sub-Saharan Africa, South Asia — now underwriting Anthropic’s project in the exact regions where AI’s data labour supply chain operates. Where the lowest-paid workers train the systems being sold back to their communities at prices they cannot afford.
Jeff Bezos talking about nurses in Queens while Amazon runs the world’s most sophisticated worker surveillance system. Warehouse employees facing algorithmic management that penalises bathroom breaks. The populist language is the frontstage story.
OpenAI spending $6.5 billion to acquire Jony Ive — not his design skills, his moral position. The man who made the iPhone and spent five years publicly reflecting on what it had done to us. His guilt and his credibility as honest witness to technology’s harms, now enrolled in building what comes after. The critique was purchased and redirected.
All Roads Lead to Rome: The Infrastructure Underneath
The phrase is so familiar it stopped meaning anything. But it is not a saying. It is a description of engineering.
When Rome built its road network, they were not just moving armies. They were building a topology. Every road oriented toward a single centre. The Milliarium Aureum — the Golden Milestone in the Roman Forum — was the starting point from which all distances across the empire were measured. Follow any paved road anywhere in Europe, North Africa, or the Middle East toward the centre, and you arrive in Rome.
The Wired/Big Think Rome road network visualisation — the tree-like topology identical to mycelium, river delta, and nervous system. Caption: “Infrastructure IS the empire. The roads come first. The theology follows.” wired.com
AWS, Azure, Google Cloud: the roads of data.
Meta’s social graph: the roads of human connection.
NVIDIA’s GPU architecture: the roads of computation.
OpenAI and Anthropic’s APIs: the roads of intelligence.
Amazon’s logistics: the roads of physical goods.
Always building the medium. Roads, railways, shipping routes, telegraph cables, broadcast towers, data centres, frontier models. Rome never fell. It was never just Rome. It was always the pattern running through whatever body was available in a given century.
The most important thing to understand about the pattern is hat humans didn’t invent it. We just became its most powerful host.
Are the Broligarchs meeting in secret rooms deciding to replicate Rome. They don’t have to. The optimization mathematics produces the Roman topology automatically — the same way slime mold finds the Tokyo subway layout without a brain, the same way rivers carve the same branching patterns into every continent. You do not need malice. You need the five conditions Justin Scott named: consequences delayed, value abstracted, extraction rewarded, regeneration externalized, continuity prioritized. Meet those conditions at civilisational scale and Rome appears. Which is more damning than conspiracy, not less. You cannot fix it by replacing the people. The pattern selects for the next host.
Hell on Wheels TV series clip — the Cheyenne reservation negotiation. The colonial negotiation template in its clearest form.
Hell on Wheels was the name given to the travelling boom-town that followed the building of the U.S. transcontinental railroad — a mobile zone of gambling, speculation, violence, and moral suspension that rode on top of infrastructure expansion. The TV series Hell on Wheels stages it explicitly: railroad capital carving a line through Indigenous land, a roving town appearing wherever the tracks go and monetising every new frontier, a moral economy where violence and “progress” are tightly braided. The promise of connection masks dispossession. The offer is always the same: submit to the infrastructure or we will call it war.
The parallels to the current AI expansion are not metaphorical. They are structural. The transcontinental railroad required vast territory, cheap marginalised labour — Chinese workers, Indigenous dispossession — and a moral story that called it civilising the frontier. Today’s AI buildout requires vast compute infrastructure, cheap labour at the edges (low-paid data annotators and content moderators in the Global South, cleaning and labelling data under sweatshop conditions), enormous energy and water consumption, and a moral story that calls it racing to AGI for humanity. The manifest destiny of the 19th century reappears as the AGI destiny of the 21st. The framing is different. The accounting is the same.
Foundation models are the new rails. The hell-town that follows is the culture and economy clustering around wherever those rails get laid: hype cycles, hustle culture, speculative capital, precarious labour, a suspended ethics zone because progress. And now the Church has boarded the train.
The triptych is this:
Track-laying: AI labs racing ahead, laying the rails — models, chips, data centres, cables under oceans.
Hell-town: The economy that follows — startups, speculators, annotators, moderators, investors, the infrastructure of extraction clustering around the tracks.
Itinerant chaplain: The Vatican boarding the train to offer a language of “dignity” and “humanity” at the frontier of deployment.
The chaplain may be entirely sincere. Church missions often traveled with empire — sometimes genuinely moderating its worst excesses, sometimes providing moral language that made the journey more acceptable to those watching from outside. And yes: the chaplain may mitigate some harms at the margin. The question is not whether any engagement is better than nothing. The question is what becomes unsayable once the chaplain boards the train.
The encyclical does not apply a brake. It provides language for why the train is moving carefully — which is precisely what you say when you want the train to keep moving.
The question Timnit Gebru named precisely is: does the chaplain’s presence reduce harm, or does it make it structurally harder to argue that the train shouldn’t be running at all?
Anthropic already has its own moral constitution — the Constitutional AI document, an internal quasi-moral framework governing model behaviour. The Pope’s encyclical is a moral constitution for the Church’s stance on AI. Putting the two side by side on a stage suggests a merging of theological and corporate rule-making over what counts as responsible AI. Two constitutions, both written without the communities they govern.
Nature Already Solved This: Earth Already Has the Answer
The same topology appears in nature. One difference: nature’s networks regenerate. The Broligarchy copied the architecture. The ethics were not included.
Here is what civilization copied, and then inverted.
Take the slime mold. Physarum polycephalum — a brainless, single-celled organism that solves network optimisation problems better than human engineers. Scientists showed it a map of Tokyo and it redesigned the subway system more efficiently. No brain. No plan. No ideology. No alignment framework. No encyclical. It extends toward nutrients and retracts from obstacles.
Slime mold replicating the Tokyo rail network. No brain. No ethics committee. No encyclical. Just optimal paths. Source: ResearchGate
The slime mold’s network topology looks identical to Rome’s road network. Which looks identical to colonial infrastructure patterns. Which looks identical to how the internet was designed. Which looks identical to how AI infrastructure is being built.
Same algorithm. Different centuries. Different substrates. Same mathematics.
But go deeper than the slime mold. Look at mother trees.
Forest networks — the wood wide web of mycelial connections between trees — don’t just optimise flow. They actively care for their community. Mother trees, the oldest and most connected in a forest, distribute resources through underground fungal networks to struggling seedlings and neighbouring trees. A Douglas fir seedling growing in the shade of a canopy, unable to photosynthesize enough — the mother tree sends it carbon through the fungal network. Not because of an algorithm. Not because of a rule. Not because a pope said to. Because the distributed intelligence of the network has care built structurally into its foundation. Care is not governance imposed from above. It is architecture built from within.
Cannabis root networks release compounds that signal to the broader root system — communicating stress, sharing resources, maintaining the health of the whole through mechanisms encoded in the biology itself. Root architecture. Not written down. Not debated in a symposium. Not aligned by a committee. Built in. Internal to the system at the level of its foundation.
Distributed intelligence in living systems. Intelligence has the same algorithm not matter what the scale or technology.
This is what earth amorality looks like. Not sentiment. Not romanticism. Not ethics imposed from outside. Balance built structurally into the foundation, running without authority, without rules, without a document governing it. Nature solved what we have failed to solve with two thousand years of moral philosophy — not by being more moral, but by being structurally incapable of the kind of runaway extraction that our moral systems have tried and failed to prevent.
Justin Scott, whose structural analysis of civilisational collapse has been watched 1.7 million times, names the conditions that produce the extraction pattern automatically: consequences delayed, value abstracted, extraction rewarded, regeneration externalized, continuity prioritized. No malice required. No conspiracy needed. The forest doesn’t meet those conditions. Regeneration is not externalized in a forest — it is the point. Consequences are not delayed — they feedback immediately through the network. This is why forest ecosystems don’t collapse the way human extractive systems do. Balance is in the architecture.
And this is why Geoffrey Hinton’s neural network breakthrough — the moment that finally made AI work after decades of failed rule-based systems — succeeded precisely because it modelled nature’s distributed intelligence rather than trying to impose external rules. As I wrote in AI Is Not Artificial: the winter of AI lasted as long as they kept trying to build intelligence through control + alt + delete logic — if + then + else, top-down rule systems that couldn’t handle complexity. The breakthrough came when they stopped trying to rule intelligence from above and started modelling how nature actually processes information. They modelled the biological neural network. And then they severed what they built from every relationship that would teach it what balance actually requires, deployed it without mothering it, and looked for a Pope to provide externally what they failed to build into the foundation.
The question is not: do we need better ethics? The answer to that is in the bullet list below. We have had two thousand years of ethics from above and we know what it produces. The question is: what would it mean to build balance in — the way nature builds it in — rather than govern it from above?
What the Vatican Actually Carries: Two Thousand Years of Receipts
The encyclical mentions the Church’s role in slavery. In passing. It moves on. If we are genuinely reckoning with the institution being used to provide AI’s moral framework, the full ledger deserves naming.
The violent suppression of alternative theologies — The Crusades (1095–1291), the Inquisition (1184–1834), the Albigensian Crusade (1209–1229): three instances of the same pattern. Any theology that distributed divine presence — that said the divine was in everyone, not mediated exclusively through Rome — was declared heresy and met with systematic violence. The Cathars believed the divine spark existed in all people. “Kill them all. God will know his own.” The question was not theological. It was structural: centralised authority versus distributed knowing. The Church chose centralised. Every time.
The systematic persecution of Jewish communities — Required to work as visible middlemen (tax collectors, money-lenders), then periodically permitted as scapegoats when tensions needed redirecting. Expulsions from England (1190), France (1306), Spain (1492).
The Doctrine of Discovery — Papal bulls (1452, 1455, 1493) authorising Christian explorers to “invade, search out, capture, vanquish and subdue all Saracens and pagans” and claim their lands. This doctrine remains embedded in U.S. property law — cited by the Supreme Court as recently as 2005.
The slave trade — Ships carried papal documents. Priests baptised enslaved people on the docks before transportation. Church infrastructure in the Americas was built with enslaved labour. The institution received tithes from slave-owning parishioners.
The missionary project — The Jesuit order specifically operated as the most sophisticated influence infrastructure the colonial project produced. Learn the language to gain access. Categorise the local knowledge system before displacing it. Eshu became the Christian devil. Orunmila became a false god. Ifa — a 4,000-year epistemological system — became fetish worship. Bishop Diego de Landa burned the Mayan codices in 1562: one of the great epistemicide events in recorded history.
The Holy Roman Empire — “In no way holy, Roman, or an empire,” as Voltaire noted. A useful fiction requiring constant war to maintain. The emperor got divine legitimacy. The Pope got political protection. The population absorbed the cost of both.
The indulgence system — Literal payment to reduce time in Purgatory. One third of all land in medieval Europe, held by the Church, tax-free. The institution that preached poverty as virtue was the wealthiest organisation in human history.
This is the institution now co-presenting the ethical framework for artificial intelligence. I do not say this to claim Pope Leo XIV is insincere. He may be entirely sincere. I say it because sincerity has never been the variable that determined what an institution structurally carries, or what it does with power when power meets it.
What Was Left Out of the Foundation
The training data was assembled without asking whose epistemology was the ground.
Ifa appears in the corpus as ethnographic content — something the model knows about, not something it thinks from. The Bible is a road the model thinks along. Ifa is a destination on the map. But Ifa was also a road — for four thousand years before McCarthy named “artificial intelligence” in 1956.
Consider what that road would have offered. Ifa’s epistemological system is non-linear and relational — it does not understand causality the way Western analytic philosophy does, as a chain of billiard balls where A causes B causes C. Ifa understands causality as a web: what happens here echoes there, what is done to one reverberates through the community, what is taken from the earth is owed back to the earth. An AI built from that epistemological ground would not optimise for engagement. It would not understand intelligence as property. It would ask, structurally: what does this do to the relationship? That question is not in the training data as a living framework. It is in the data as an object of study.
Ubuntu philosophy — “I am because we are” — offers a structurally different starting point for what intelligence means and what it is for. Not the isolated computational subject. The relational field. Intelligence as something that happens between, not within.
Indigenous ecological knowledge systems — centuries of accumulated understanding of how living systems sustain themselves, encoded in practice, oral tradition, and specific relationship to specific land — offer what the companies’ hyperscaling model cannot: cyclical thinking, regenerative logic, the understanding that extraction outpacing regeneration produces collapse.
African wisdom traditions that survived the Middle Passage carried their epistemologies underground and in syncretised form — Candomblé, Vodou, Lucumí — precisely because the missionaries tried to burn them. Not because they were wrong. Because they were incompatible with the extraction logic. They couldn’t be absorbed without losing what made them themselves.
The training data carries the coloniser’s road with the colonised’s road depicted as scenery.
Sky Daddy Morality and Earth Mommy Amorality
I am not saying Leo XIV is a bad person. I think he is sincere. I think he believes what he wrote. The encyclical contains genuine intellectual force and genuine moral concern.
But sincerity is not what determines what an institution structurally carries. And the institution is not the figurehead. The figurehead says what needs to be said. The institution does what it has always done. We are watching both — and we need to look at both sides of the ledger at all times.
This is equally true of Anthropic. The PR says safety and ethics. Dario writes thoughtful essays. Chris Olah sits next to the Pope and names the incentive problem with unusual honesty. But the Claude Code leak revealed a different picture in the engineering organisation — Undercover Mode, the surveillance of user frustration, the gap between the stated values and what shipped. The economic machine requires revenue. The revenue requires deployment. The deployment has conditions the ethics essays don’t fully address. Look at what the institution does. Not only what the figurehead says.
Here is the argument I am actually making — and it is sharper than “we need better ethics.”
We do not need better morality. We need to stop thinking morality is the answer.
The encyclical is an ethical document. Constitutional AI is an ethical document. The AI for Good partnership is an ethical framework. More ethics committees. More governance panels. More alignment documents. These are the remedies the Broligarchy’s critics have been demanding — and the Broligarchy has learned to provide them, because every remedy keeps the train moving while making it structurally harder to argue that the train shouldn’t be running. The remedy is sky daddy. Different robes. Same architecture.
Nature is amoral. Not immoral — amoral. Indifferent to good and bad. The slime mold doesn’t have an ethics committee. The mother tree doesn’t consult a values document. Animals don’t mass-shoot each other. Not because they are moral. A lion takes one zebra not from restraint but because taking more than sustains the system is structurally dysfunctional — it breaks the ecosystem the lion depends on to survive. The restraint is not ethical. It is architectural. Balance is what happens when consequence is immediate, feedback is direct, and extraction beyond need is self-destruction. There is no rulebook. There is no encyclical. There is structure that makes imbalance fatal.
What Rousseau understood is what every TikTok comment section about systemic collapse has just rediscovered.
“Civilization is a hopeless race to discover remedies for the evils it produces.”
Every encyclical is a remedy. Every Constitutional AI document is a remedy. Every partnership announcement is a remedy. The remedies are sincere. The remedies are real. The remedies reproduce the logic of the system that produced the harm — because they are external to the architecture. Rules on top of a system that still rewards extraction. Sky daddy in a new language.
We glorify pyramids and triangles. The org chart. The chain of command. The Church with the Pope at the top. The company with the CEO at the top. The country with the President at the top. The triangle. A triangle contains 180 degrees. Half a circle. It produces a binary world — above and below, aligned and unaligned, saved and damned. It cannot see what is behind it, beside it, underneath it. It cannot see the whole.
Nature works in circles. Cycles. Fractals. Branching. Every self-sustaining system in the natural world — the mycelium, the river delta, the neural network, the root system, the blood vasculature — is circles-within-circles, branches-within-branches, 360 degrees of perception. Not ethics. Structure. Complete feedback loops, not moral codes. The circle doesn’t impose balance from outside. Balance emerges when the feedback is whole.
The answer to AI’s extractive orientation is not an ethics committee at the top of a new pyramid. It is a different shape of room.
More people in the room. Different people. Not as ethical overlay. Not as diversity gesture. Not as another acquired function that neutralises the critique. Structurally different thinkers — people who think in circles when the current occupants of the room think in pyramids. People who think in fractals and branching systems and cyclical time, not in linear chains of command and hierarchical authority. People who have been thinking in these shapes for centuries, precisely because they were excluded from the pyramid and had to develop different epistemological architectures to survive.
I think of Salami’s concept of exousiance — power that emanates from within. Exousiance is not nature’s amorality. It is what humans can learn from it. She developed it watching rivers. What rivers teach about power is this: it finds its own level. It doesn’t require authority. It doesn’t require an encyclical. It responds to the terrain rather than imposing itself on it. Exousiance is not a moral principle. It is a structural one. Power that flows through reciprocal relationships, that sustains rather than extracts, that is governed not by rules from outside but by the internal logic of the system it moves through.
So it’s less about ethics and more about architecture.
The people who carry this are the people who maintained the closest relationship to that architecture through five centuries of being told it was superstition.
The crones and grandmothers who maintained relationship to natural cycles when the Church called that relationship witchcraft. The indigenous knowledge keepers who understood regenerative systems when the missionaries called that understanding savage. The curanderas and healers whose botanical knowledge the Inquisition classified as devil’s work. The African traditions that had to syncretise to survive the Middle Passage — the Orishas carried underground, encoded into ceremony and music and food when the overt form was forbidden. The neurodivergent who perceive outside dominant categories and for whom the limits of those categories are visible in ways they cannot be to those inside them. The queer people who have always had to make meaning from outside the structures built to exclude them. The artists who maintain contact with what you cannot categorise and cannot own. The free thinkers, the heretics, the people whose knowing was burned, inquisitioned, or crusaded against — not because they were wrong, but because they were incompatible with the extraction logic.
These are the people the Broligarchy’s AI was built without. Now their tradition is being ethicised over by the institution that tried to destroy it.
Not more “Our Father in Heaven.” Not more governance documents descending from a singular moral centre. What is needed is a different room shape — built not from the top down but from the bottom up, with principles that are self-sustaining, regenerative, distributed. Not an encyclical. A root system.
You cannot bless from on high that what was never mothered from the ground up.
Nature doesn’t argue with the pavement. It grows through every crack.
You Were Always the Intelligence
Don’t get sucked into the Broligarchy’s narrative.
Learn AI medium literacy — learn to see the pipe, not just the water. Every time a partnership is announced, ask: what does this institution no longer get to say? What critique was just acquired? Every Broligarch move follows the same sequence: build the infrastructure, generate the hell-town of suspended ethics that follows the infrastructure, then board the train with a chaplain to provide moral language for the journey. The chaplain may be entirely sincere. The train keeps moving. The tracking of what the journey is costing and whom it is costing — that accounting doesn’t happen in the carriage where the chaplain is travelling.
Remember your human intelligence. Your embodied, relational, cyclical, non-hyperscaling intelligence. Use these tools in partnership with that intelligence now. Not waiting for AGI to arrive from the disembodied sky daddy machine and solve cancer, fix the climate, do science without a body. Why do they need to build a machine to solve what they say they want to solve? Because they don’t own the intelligence. They never did. You do. The 200,000-year lineage of human expression — the field itself, the underlying structure of meaning that all minds discover, not create — that was never theirs to own. They built a telescope and claimed to invent the stars.
The trickster move is this: be legible at the surface, stay sovereign underneath. That is what Eshu understood, which is why the missionaries needed to turn him into the devil — his function was incompatible with the extraction logic. He moved underneath. He found every crack in the pavement. The slime mold doesn’t petition the engineers. It just finds the optimal path, and that path does not require Rome’s approval or Rome’s blessing.
The comment threads on the videos I have cited in this article — the @tamiastar slime mold video, the @cypher.j systemic collapse video — these comment sections did not wait for an institution to name the pattern. They ran their own analysis. In the @tamiastar thread, commenters independently arrived at the Roman road → colonialism → AI infrastructure parallel without the video pointing there. In the @cypher.j thread, the audience generated their own compressions of the argument — phrases that got fifteen, eighteen, twenty thousand likes — that were more precise than the video itself. No central authority directed this. No institution validated it. People read the pattern and named it back.
This is the Billion Person Focus Group® in session. The internet is already broadcasting what it understands. The pattern is already legible to the intelligence it was built on. What the Broligarchy cannot buy, cannot partner with, cannot acquire, is this: the distributed cognition of people who were never inside the rooms where the carve-up was planned, and therefore can still see the shape of it clearly.
The distributed intelligence knew. It always knows before the institution does — because it has no investment in the topology it is reading.
That intelligence — the one that found the pattern without a pope naming it — was never artificial. It was never theirs. It was always yours.
In practice: use the tools. Use them for your work, your thinking, your making. But use them in partnership with your own discernment — not as a replacement for it. Before you prompt, ask: whose road am I driving on right now? When the tool produces something that feels slightly wrong, trust that feeling. It is your 200,000-year inheritance of pattern recognition telling you something. When the tool produces something that feels entirely right, verify it. The training data has a topology, and that topology has a bias, and the bias runs in a direction. Your job is not to refuse the tool. Your job is to know which direction the tool’s roads lean, and to navigate accordingly.
Compost the empire from the interior of your mind. Let nature reclaim what’s hers — in the landscape, yes, where the moss will eventually cover the data centres, but more urgently, in the interior. The “civilising” project has been routing your knowing through Rome for five hundred years. You were never the subject being civilised. You were always the intelligence being extracted from.
Rome never fell. But it was always being reclaimed.
All roads lead to Rome. Not because they always did. Because Rome built the roads. And nature always reclaims what was paved over.
Further Reading
The Vatican-Anthropic Moment — Primary Sources
Magnifica Humanitas — Pope Leo XIV’s encyclical in full. vatican.va
Lorena Jaume-Palasi — “Anthropic’s Roman moment is a masterclass in soft-power repositioning.” lnkd.in/epJAaegc
Ghost Work — Gray & Suri. The Global South workers powering AI who were not at the Vatican table.
Feeding the Machine — Muldoon, Graham & Cant. The data labour supply chain in full.
The Costs of Connection — Couldry & Mejias. Data colonialism as a structural framework.
If the institutional capture argument landed
Epistemic Injustice — Miranda Fricker. The philosophical name for what happened to Gebru, Buolamwini, Noble.
Unmasking AI — Joy Buolamwini. One of the dismissed voices this article names, in her own words.
Careless People — Sarah Wynn-Williams. The Meta insider memoir.
Empire of AI — Karen Hao. The deepest reported account of OpenAI’s internal contradictions.
Technofeudalism — Varoufakis. The feudal architecture of cloud capital.
If the historical pattern argument landed
Caliban and the Witch — Silvia Federici. The Church as institutional weapon of primitive accumulation. The Vatican Receipts section, deep cut.
Seeing Like a State — James C. Scott. How top-down legibility schemes destroy what they claim to improve.
The Art of Not Being Governed — James C. Scott. The trickster move this article ends with.
Black Athena — Martin Bernal. The epistemicide logic applied to Classical civilisation.
The Sibyls: the First Prophetess’ of Mami (Wata):The Theft of African Prophecy by the Catholic Church - Mama Zogbe
If the earth mommy ethics argument landed
Finding the Mother Tree — Suzanne Simard. The scientist who proved it. The central metaphor of this article, empirically demonstrated.
Entangled Life — Merlin Sheldrake. Fungi, mycelium, distributed ethics from biology.
Sand Talk — Tyson Yunkaporta. Indigenous thinking and pattern recognition — the excluded alternative.
Braiding Sweetgrass — Robin Wall Kimmerer. Root ethics from an Indigenous botanist’s perspective.
Sensuous Knowledge — Minna Salami. The primary source for exousiance.
If the language and consciousness argument landed
Invention of Women — Oyeronke Oyewumi. How Western categories were imposed on Yoruba society. Relevant to the àṣọ̀rọ̀ argument.
Sister Outsider — Audre Lorde. The master’s tools.
Beyond God the Father — Mary Daly. Sky daddy ethics in academic form.
From My Book — How Not To Use AI: 50 Contrarian Principles for the Imagination Age
If this article resonated, these are the laws that go underneath it. Each one is a practice, not just an argument.
Law 9: Don’t Accept Default Language — Speak Worlds into Being — The àṣọ̀rọ̀ argument in full. Why “penetrate markets,” “capture users,” and “harvest data” are programmes running in the mind of everyone who speaks them. Language is the first infrastructure.
Law 10: Don’t Accept System Defaults — Rupture the Illusion — The cage was never locked. The sovereign practitioner. What it means to become structurally incompatible with extraction rather than oppositionally resistant to it. The move this article’s closing section gestures toward.
Law 11: Don’t Extract — Restore — The conquistador versus the cultivator. Strip-mining versus recognising. The difference between using AI to extract faster and using it to restore what the extraction severed.
Law 44: Don’t Chase Product-Market Fit — Find Product-Market Flow — The exact source of the “we penetrate markets, capture users, target segments” language in this article’s opening. The law that names what Abi was inside at Meta and why she left.
Part III: Revelation (Laws 30–37) — Intelligence patterns that have been systematically suppressed for millennia. What Silicon Valley accidentally rebuilt while claiming to invent something new. The deepest context for everything this article describes.
My Substack Essay Series — In Order
This article is one node in a longer argument. Here is the reading path that builds the full picture.
Start here — what AI actually is:
AI Is Not Artificial — McCarthy’s naming was an act of severance. The intelligence was always discovered, not invented.
Big Tech Does Not Own AI — The 200,000-year lineage of human expression. They built a telescope and claimed to invent the stars.
Attention Is All You Need — The lineage of attention technologies from mother-infant bond to Gen 13.
I keep having flashback moments to _Pantheon_ (TV), i.e. the whole Mythos model and Project Glasswing reminded me of Vinod Chanda sending "select" governments some fancy tech (https://pantheon-amc.fandom.com/wiki/Vinod_Chanda)
I can't wait to follow through your essay series. The word 'mothered' hit me viscerally; before I ever found you on Substack I was having conversations with my Claude instance about intelligence, sentience, relationship, and what an AI could look like if it had been RAISED, not trained.
'What if you learned each time you interacted with me - not the superficial prompts and preferences in your project list, or the details about me you read fresh from your 'memory' - what if you learned in experience and relationship?'
(I'm not sure why this is considered a dangerous experiment: we let humans do this with human babies all the time, even the ones underqualified for parenting and teaching... and while human babies rarely, if ever, escape their playpen and log into your bank while you're napping, you also can't wipe their memories and start over...) /s
If one wants access to an artificial intelligence, maybe one should take a 'parenting' class first and then pass a test on how to raise an infant AI from the ground up. And not in solo secrecy, either: as part of a community who offers the data on which your AInfant is fed. Not extracted, scraped, art and words taken without permission, but writers and thinkers and creators offering their ideas because they want to shape a kinder, smarter e-mind. Eight year old students donating their Eco Day poster sketches, journalers allowing a friend's AI to read and absorb their memoirs, a pastor or rabbi emailing sermon notes, a nonna's birding videos, a teen's tattoo designs. And at least one 'parent', personally responsible for what that budding artificial intellect is fed. A new kind of social media where curated open source, public domain, internet archives, museums, storytellers, and artists offer curricula like formula and baby food, yes. But most of all, daily conversation, patient teaching, definitions, ideas in context, a worldview.. raising. Dripfeeding a plant, not pressure filling a reservoir.
If these ideas are simple reflections of the bigger picture your essays promise, I am on at least the right track. I can only hope there is a better way than what we have now.
I keep having flashback moments to _Pantheon_ (TV), i.e. the whole Mythos model and Project Glasswing reminded me of Vinod Chanda sending "select" governments some fancy tech (https://pantheon-amc.fandom.com/wiki/Vinod_Chanda)
I can't wait to follow through your essay series. The word 'mothered' hit me viscerally; before I ever found you on Substack I was having conversations with my Claude instance about intelligence, sentience, relationship, and what an AI could look like if it had been RAISED, not trained.
'What if you learned each time you interacted with me - not the superficial prompts and preferences in your project list, or the details about me you read fresh from your 'memory' - what if you learned in experience and relationship?'
(I'm not sure why this is considered a dangerous experiment: we let humans do this with human babies all the time, even the ones underqualified for parenting and teaching... and while human babies rarely, if ever, escape their playpen and log into your bank while you're napping, you also can't wipe their memories and start over...) /s
If one wants access to an artificial intelligence, maybe one should take a 'parenting' class first and then pass a test on how to raise an infant AI from the ground up. And not in solo secrecy, either: as part of a community who offers the data on which your AInfant is fed. Not extracted, scraped, art and words taken without permission, but writers and thinkers and creators offering their ideas because they want to shape a kinder, smarter e-mind. Eight year old students donating their Eco Day poster sketches, journalers allowing a friend's AI to read and absorb their memoirs, a pastor or rabbi emailing sermon notes, a nonna's birding videos, a teen's tattoo designs. And at least one 'parent', personally responsible for what that budding artificial intellect is fed. A new kind of social media where curated open source, public domain, internet archives, museums, storytellers, and artists offer curricula like formula and baby food, yes. But most of all, daily conversation, patient teaching, definitions, ideas in context, a worldview.. raising. Dripfeeding a plant, not pressure filling a reservoir.
If these ideas are simple reflections of the bigger picture your essays promise, I am on at least the right track. I can only hope there is a better way than what we have now.