From Rerum Novarum to Magnifica Humanitas: Why 2028 Is Our Nehemiah Moment
The Echo Is Too Loud to Ignore
Some déjà vu is comforting. This isn't that kind.
In 1891, steam was king. Factories had bulldozed the old artisan guilds. Children worked 14-hour shifts. A new class of industrial barons amassed wealth so quickly it made governments nervous, while the workers operating the machines lived in conditions we'd now classify as humanitarian crises.
Society was coming apart at the seams. And almost nobody with institutional authority had the credibility, or the nerve, to say so out loud.
Then a very old, very global institution stepped up. Pope Leo XIII published an open letter called Rerum Novarum, "On New Things." It mapped a third path between unfettered capitalism and revolutionary socialism.
It didn't fix the Industrial Revolution overnight. It planted the seeds: a century of labor law, workers' rights, and middle-class expansion we now treat as background scenery.
Fast forward 135 years, almost to the day. The current head of that same institution, Pope Leo XIV, signed a sequel: Magnifica Humanitas, "Magnificent Humanity." Signed May 15, 2026. Released May 25.
That symmetry wasn't an accident.
At the presentation, the framing was blunt: "Today we find ourselves facing a transformation of similar magnitude, with perhaps even greater consequences. Artificial intelligence already touches many areas of our lives and affects decisions that shape human coexistence."
Translation: one of the oldest continuously running organizations on Earth just pulled the fire alarm and pointed at the same systemic disruption it flagged 135 years ago.
History doesn't repeat. But when it rhymes this loudly, only a fool keeps the earbuds in.
The Industrial Reckoning: What History Actually Got Right
Forget the incense for a second. Read Rerum Novarum as a systems document.
By 1891, the Industrial Revolution had been underway for roughly 60 years in Britain and was spreading rapidly across Europe and America. The guild system that structured skilled labor for centuries had collapsed. Rural populations flooded the cities. Capital was winning. Decisively. At labor's expense.
The 1891 document refused to pick the side everyone expected. It rejected abolishing private property, calling it economically broken and socially destructive. It also rejected the laissez-faire fairy tale that markets magically produce fair outcomes on their own.
Instead, it said something radical for the era: the dignity of human work is not contingent on its raw market value.
Here's the part that should make your hair stand up on end. Map the 1891 principles against 2026, and they line up like a horror-movie sequel.
Dignity of Labor:
- 1891: Workers treated as human beings, not machine inputs.
- 2026: Knowledge workers are being systematically replaced or deskilled by AI systems.
The Economic Baseline:
- 1891: Pay sufficient to sustain a stable, dignified life.
- 2026: AI's economic gains are pooling with capital owners, not the workforce.
Local Autonomy (Subsidiarity):
- 1891: Local institutions handle what they can, instead of over-centralizing.
- 2026: Decisions about a community's AI infrastructure are made by distant tech corporations with zero local input.
Right of Association:
- 1891: Workers earned the right to organize unions.
- 2026: The urgent need for democratic oversight bodies and civil-society voices in AI governance.
The Public Sector's Role:
- 1891: Governments must protect workers' fundamental rights.
- 2026: Governments must ensure AI accountability, not hand the keys to private tech firms.
Same disease. New century. Better Wi-Fi.
Now, the optimistic reading, because it's real: the Industrial Revolution, for all its misery, eventually raised living standards, grew the middle class, and spread productivity gains more fairly. The world adapted.
But notice the fine print on that timeline. 50 to 70 years of painful adjustment.
When the suffering spans two generations, "it worked out eventually" is a sentence you can only say if you weren't there for it.
The question for 2026 isn't whether we can put the genie back in the bottle. It's whether we can shorten the pain phase. Steer it, don't bottle it.
Magnifica Humanitas: Updating the Playbook for the Cognitive Age
Rerum Novarum addressed the replacement of muscle by machines.
Magnifica Humanitas deals with something weirder: machines replacing thought itself.
The document runs 42,300 words and was signed on the exact same calendar date as its predecessor. That's not a coincidence. That's a citation. This isn't a sermon. It's a civilization-level critique using the same intellectual architecture that built a century of labor rights.
The central allegory is the Tower of Babel. Strip the religion out, and it's a story about what happens when raw technical capability sprints way ahead of collective wisdom and communication.
The diagnosis: we're building those towers again. Tall, impressive, and missing the shared infrastructure required to actually govern them.
Here's the short list of warnings, in plain English.
On Human Agency and Bias: AI systems have no conscience and no lived body. They can't actually understand what human means. They're also increasingly the ones deciding who gets healthcare, a job, or a loan, "on the basis of data tainted by prejudice and injustice." This isn't hypothetical. It's already documented in mortgage lending, criminal sentencing, and healthcare resource allocation.
Reality Check: A model trained on biased history doesn't fix the bias. It launders it into math, then hands you a confident decision with no fingerprints.
On Work and Social Cohesion: The document echoes what Columbia University researchers have quantified: AI adoption is driving declines in labor share. Work isn't just a paycheck. Its identity, routine, and belonging. Kill work too fast, and you don't just cut income; you cut jobs. You delete a load-bearing wall of human purpose.
On Power Concentration: This is the sharpest jab. It's the risk of technocratic concentration: a tiny class of AI developers, financiers, and deployers hoarding decision-making power at a scale that quietly hollows out democracy.
The kicker? Anthropic co-founder Christopher Olah attended the public presentation. One of AI's most safety-obsessed builders stood there and admitted you need "people outside those incentives — people who care about things going well and insist on safety."
When the people building the thing are asking for adult supervision, that's not modesty. That's a warning label.
On the Truth Ecology: Synthetic disinformation isn't a bug. In some applications, it's the product. The document frames this as an attack on the shared reality that free societies need to function. Generate convincing lies at scale, and the commons of shared truth starts to rot.
The Disarmament Imperative: The text literally calls for AI to be "disarmed." Strong word, chosen on purpose. The parallel is nuclear arms control: nobody's asking to delete the underlying science. They're demanding the same grim seriousness humanity eventually applied to weapons capable of destroying cities.
One thing worth nailing down: Magnifica Humanitas is not anti-AI. It's anti-recklessness. The argument is simple. Technology isn't the enemy. The open question is whether it serves human flourishing or becomes another tower from which nobody can climb down.
Why 2028 Is a Narrow Strategic Window
Here's where it gets concrete. And uncomfortable.
The Industrial Revolution took roughly 60 years to go from the first factories to the establishment of real institutional frameworks. AI is moving approximately 27 times faster, according to adoption metrics. What steam did to manufacturing over decades, AI is doing to knowledge work in months.
The Stanford HAI 2026 AI Index lays it out. Frontier models gained 30 percentage points in a single year on Humanity's Last Exam, a benchmark deliberately built to be brutally hard for AI and friendly to human experts.
Quick translation for the non-nerds: a benchmark is a standardized test for machines. We write these exams specifically to be hard so we can measure how smart the models are getting. The catch? The models are now taking the exams faster than we can write new ones. Tests meant to last years are getting solved in months.
It gets better. On OSWorld, which tests AI agents on real computer tasks (clicking, typing, navigating apps like a person would), accuracy jumped from about 12% to 66.3% inside a single measurement window. That puts the machines within six percentage points of the human baseline.
An "agent," by the way, is just an AI that doesn't only chat. It does things. Opens your browser, fills the form, and books the flight. Picture an intern who never sleeps, never asks for clarification, and occasionally books the flight to the wrong continent with total confidence.
The people building this stuff aren't whispering about timelines. They're shouting.
Anthropic co-founder Jack Clark puts the odds at 60%+ that AI systems capable of autonomously building their own successors, with zero humans in the R&D loop, will exist by the end of 2028.
Sit with that. Machines are designing better machines, with us watching from the lobby.
Sam Altman went further, suggesting that by 2028, "most of humanity's intellectual capacity could reside inside data centers rather than outside them."
These aren't doomsday cranks. These are mainstream projections from the founders themselves.
Which brings us to a guy from 445 BCE.
Nehemiah was a high-ranking official who got word that Jerusalem's defensive walls were in total ruin. The people were exposed, vulnerable, unable to function as a community. He didn't panic. He didn't file the problem under "abstract." He secured resources, traveled to the site, organized the population, and finished a rebuild that had been stalled for decades.
In 52 days.
The takeaway isn't divine intervention. It's operational organization. He surveyed the damage at night before announcing a thing. He assigned specific crews to specific work. He kept people defended while they built.
The 2028 window is our 52 days. Not because the world ends after it. Because the governance structures, industry norms, regulatory frameworks, and accountability mechanisms we build now, before autonomous AI R&D becomes standard, get exponentially harder to bolt on later.
What makes this moment different:
- Compressed Speed: The capability curve is measured in months, not decades.
- Cognitive Scope: This reshapes knowledge work (law, medicine, governance, education), not just physical labor.
- Power Concentration: A handful of private labs in a few cities control the most consequential technology since nuclear fission.
- Infrastructure Strain: AI data centers are already pushing up utility bills for ordinary people across the country.
The middle path here isn't between cheerleading and rejection. It's between naive acceleration and counterproductive panic. Pick neither. Pick the boring third option that actually works.
The Rising Backlash: Predictable, Valid, and Navigable
Here's something the tech optimists keep underestimating: the backlash isn't fringe anymore.
Between May 2024 and June 2025, an estimated $162 billion in U.S. data center projects got blocked or delayed by local opposition. A Heatmap News poll found 52% of Americans oppose or strongly oppose building an AI data center near where they live. By early 2026, 99 of the 770 planned data centers tracked nationwide were being actively contested in courts and local councils.
And it's escalating. In Wisconsin, residents used a ballot measure to block a proposed AI facility. In Indiana, a local official who backed a data center project faced political violence. In early 2026, an individual targeted OpenAI's headquarters and executive property. On the policy front, Senators Bernie Sanders and Alexandria Ocasio-Cortez introduced the AI Data Center Moratorium Act in March 2026 to pause new construction.
Read that lineup again. Sanders and AOC on one side of the aisle, populists on the other, both pointing at unchecked AI deployment and calling it a threat to the working class. When the far left and the far right agree on something, the building is, in fact, on fire.
This is the exact pattern Rerum Novarum warned about. A fast technological shift delivering real benefits alongside severe localized harms, with those harms dumped on communities that feel completely shut out of the decision.
Quick history correction while we're here. The original Luddites get slandered constantly. They weren't terrified of technology in the abstract. They were skilled textile artisans whose livelihoods were being erased by factory owners with no legal obligation to share a cent of the productivity gains.
The machine-smashing was desperate collective bargaining in a world with no regulatory protections. They weren't anti-progress. They were anti-getting-robbed.
Today's anti-data-center protests have the same skeleton. Real concerns about energy costs, water usage, noise, and displacement, raised by citizens who feel ignored.
Here's the thing. The strategic mistake is dismissing this as technophobia. The operational mistake is meeting it with confrontation. What actually works is genuine stakeholder engagement, equitable benefit distribution, and accountability mechanisms that give communities real influence.
Boring. Effective. Pick boring.
The Proactive Fortification Toolkit: What Rebuilding Looks Like
Nehemiah's 52-day project wasn't a miracle. It was project management with better stakes.
He surveyed the damage before making public statements. He secured resources upfront. He assigned each group the section of wall nearest their own home, so the work got personal. He kept half the workforce on security while the other half built.
The modern version is just as systematic. Build the safeguards now, while there's still runway.
Here's the short list:
- Audit Before You Deploy. Any AI making high-stakes calls about people (credit, healthcare, employment, public benefits) gets audited for bias and disparate impact before rollout. Not after the lawsuits. Not after the headlines.
- Invest in Human-AI Symbiosis, Not Blind Replacement. The companies actually winning with AI aren't the ones nuking their whole workforce. They're designing workflows where AI handles the routine data grind and humans do strategy, empathy, and judgment. Pure replacement bites back: MIT research on automated manufacturing found aggressive AI adoption can cause initial productivity declines of up to 60 percentage points before things stabilize. Translation: rip out the humans too fast and your shiny new system faceplants on the way to the savings.
- Build Transparent, Accountable Systems. Every autonomous agent in a high-stakes environment needs verifiable audit trails and a clear human escalation path. "The algorithm made an unreviewable error" should never fly as a defense in law, medicine, finance, or public governance. "Computer says no" is not due process.
- Engage Communities Proactively, Not Defensively. Data center opposition is surging because people refuse to be excluded. The playbook that works: total transparency, real local benefit-sharing (tax revenue, infrastructure upgrades, energy guarantees), and agreements with teeth.
- Support Agile Governance. Traditional regulation studies a problem for years before writing a rule. AI capability curves laugh at that timeline. We need adaptive governance: standing technical advisory bodies with the authority to update safety standards dynamically as capabilities scale.
What's Coming: The Second-Order Structural Shifts
The scariest risks in 2028 aren't the sci-fi headlines. They're the quiet structural shifts nobody puts on a movie poster.
Ghost GDP: Picture AI-driven productivity gains showing up beautifully in national statistics while never reaching the actual economy. Machines don't buy homes, pay local taxes, or shop at the corner store. If the wealth pools entirely with capital owners and never recirculates into labor income, the growth charts look fantastic while the consumer economy quietly deflates underneath. Healthy vitals on the monitor. No pulse in the room.
Benchmark Saturation: A measurement crisis unique to this era. We build tests to measure AI, and the models master them faster than we can design new ones. So our ability to know where we even are on the capability curve, and when to intervene, is going blind. We're flooring an accelerating car while every gauge on the dashboard pegs to max and stops reporting.
The Guardrail Imperative: This is the whole point. Building frameworks for AI isn't about switching it off. Nuclear non-proliferation didn't kill nuclear medicine or nuclear energy. It built an international framework to stop the catastrophic misuse. Same goal here: keep powerful systems inside the moral and legal accountability structures civilized societies apply to every other major force.
The 2028 window is real. Rapid agentic scaling, compressed R&D timelines, rising political backlash, and evolving regulatory playbooks have briefly cracked the door open. After it shuts, economic and systemic path dependencies lock in, and corrections get exponentially harder.
The door is open. For now.
The Bricks Are Here
When Nehemiah surveyed those ruined walls, he didn't see a tragedy. He saw an organizational problem with a deadline.
He organized the builders. He insulated them from distraction. He kept the stakes local and tangible.
The ruin we're trying to prevent isn't physical. It's structural. It's the gap between how fast technology moves and how slow our governance crawls.
But the raw materials are sitting right here. Thoughtful policies taking shape. Major institutional voices providing historical guardrails. Internal safety researchers begging for accountability. Local communities demanding a seat at the table.
The question was never whether this transformation continues. It will. The genie's not going back in the bottle.
The real question is whether we organize the builders today, or spend the next decade mopping up the fallout because we couldn't be bothered to pick up the first brick.
History says the choice is ours. The window says choose fast.
The bricks are here. The window is narrow. Time to build.
This is part of a series exploring the intersection of AI capability, historical precedent, and institutional guardrails. Future installments will examine specific enterprise applications of proactive stabilization and an analysis of emerging AI governance frameworks across global markets.
