This week, something quietly important happened: a frontier AI researcher stood at the Vatican as Pope Leo XIV released an encyclical on artificial intelligence.
If you think that's just a symbolic photo-op, you're missing the point.
AI has moved beyond engineering. It is now a civilizational question: power, work, truth, war, and what we mean by "human" are all on the table.
The most honest line I've heard from "inside the lab"
Chris Olah (one of the leading minds in frontier AI) said something rare for a builder to admit:
Frontier AI labs operate under incentives that can conflict with doing the right thing. So the world needs outsiders—critics and institutions not financially dependent on the outcome.
That line matters because it punctures a comforting illusion: that good intentions inside top AI companies are enough.
They're not.
Not because people inside labs are immoral, but because the incentives are structurally strong:
• Ship faster
• Raise more money
• Win enterprise deals
• Beat the rival lab
• Serve geopolitical priorities
• Reduce regulation
• Control the narrative
At that scale, "ethics" cannot survive as a slide deck. It needs institutions, laws, and independent oversight.
The Pope's central claim: AI must remain subordinate to the human person
Pope Leo XIV's encyclical (Magnifica Humanitas) enters the debate with a different vocabulary—dignity, conscience, solidarity, the common good.
But it lands on a practical conclusion:
AI must serve the human person, not reorganize the human person around AI.
In other words, we cannot judge AI only by:
• Capability
• Speed
• Productivity
• Valuation
• Engagement
We must also judge it by what it does to:
• Human agency
• Work and livelihoods
• Truth and social trust
• Children and the vulnerable
• Inequality and exclusion
• Peace and the risk of war
"AI imitates intelligence, but it is not human"
This is a distinction the tech world often avoids because it feels philosophically inconvenient.
The Pope insists it matters.
AI does not have:
• A body
• Conscience
• Moral responsibility
• Lived experience
• Love, friendship, or inner growth
That's not an insult to the technology. It's a warning to us.
Because the moment we treat machines as moral agents, humans start escaping accountability.
A company can say: "the model decided."
A platform can say: "the feed optimized itself."
A military can say: "the target was selected algorithmically."
But none of those statements is morally sufficient.
Systems do not absolve people. Algorithms do not dissolve responsibility.
The real issue is power concentration
AI is not emerging in a neutral field.
The infrastructure—compute, chips, cloud, platforms, distribution, data—sits in very few hands.
When power concentrates, the risks are predictable:
• Opacity (nobody can explain what's happening)
• Dependency (societies become reliant on private systems)
• Manipulation (attention and belief can be engineered)
• Exclusion (some people become invisible)
• Inequality (benefits accrue to a small group)
This is where the Vatican's language and the investor's language meet.
Different words. Same reality.
Intelligence without accountability becomes power without restraint.
AI is not "immaterial magic"—it rests on real labor and real resources
One of the most grounded parts of the Pope's message is about work.
Behind every "instant" AI output is a physical and human supply chain:
• Data centers and energy
• Minerals and hardware supply chains
• Data labelers and trainers
• Content moderators
• Low-paid, invisible labor that makes the model look "smart"
The machine looks frictionless because the friction is pushed onto people we don't see.
Founders, investors, policymakers: this matters. The AI economy is not just software. It is labor, energy, hardware, and politics.
Truth is a civic issue, not a product issue
AI can generate content faster than societies can verify it.
That's not just "a misinformation problem."
It's a stress test for democracy and social trust.
When synthetic certainty floods public life, the boundary between truth and manipulation blurs—and people lose the shared ground needed to make collective decisions.
The hardest red line: autonomous weapons
The Pope is strongest here:
Lethal or irreversible decisions must not be delegated to opaque automated systems.
Once speed becomes the highest military virtue, human judgment becomes "the bottleneck" to remove.
That path is dangerous.
The deeper message is not anti-technology. It is anti-idolatry.
This is the part worth sitting with.
The encyclical does not reject AI.
It rejects the worship of AI.
It rejects the idea that:
• Progress is whatever increases capability
• Efficiency is the highest moral good
• Humans must justify their worth through productivity
• Private power can govern civilizational infrastructure by itself
So what do we do with this?
AI cannot be left only to the people racing to build it.
Again: not because they are bad people, but because no group should be builder, beneficiary, and regulator of a technology that reshapes work, truth, war, and human self-understanding.
We need outsiders—with real power and legitimacy:
• Scientists who can audit systems
• Regulators who understand incentives
• Workers who understand displacement
• Teachers who understand children's attention and development
• Philosophers and theologians who defend human dignity
• Investors who admit capital shapes behavior
• Citizens who refuse to outsource moral judgment
The right question is no longer "Can we build it?"
The right question is: What kind of human world are we building with it?
Here's your first step (next 24 hours)
Pick one area where you have leverage—company policy, investing criteria, governance, education, or media—and write down:
1. What is the human value we refuse to trade away?
2. What is the minimum oversight or constraint we will insist on?
3. Who are the "outsiders" we will invite in (and listen to) before we scale?
Clarity first. Speed second.
B Hari
Simplicity with substance
www.bhari.com
If you think that's just a symbolic photo-op, you're missing the point.
AI has moved beyond engineering. It is now a civilizational question: power, work, truth, war, and what we mean by "human" are all on the table.
The most honest line I've heard from "inside the lab"
Chris Olah (one of the leading minds in frontier AI) said something rare for a builder to admit:
Frontier AI labs operate under incentives that can conflict with doing the right thing. So the world needs outsiders—critics and institutions not financially dependent on the outcome.
That line matters because it punctures a comforting illusion: that good intentions inside top AI companies are enough.
They're not.
Not because people inside labs are immoral, but because the incentives are structurally strong:
• Ship faster
• Raise more money
• Win enterprise deals
• Beat the rival lab
• Serve geopolitical priorities
• Reduce regulation
• Control the narrative
At that scale, "ethics" cannot survive as a slide deck. It needs institutions, laws, and independent oversight.
The Pope's central claim: AI must remain subordinate to the human person
Pope Leo XIV's encyclical (Magnifica Humanitas) enters the debate with a different vocabulary—dignity, conscience, solidarity, the common good.
But it lands on a practical conclusion:
AI must serve the human person, not reorganize the human person around AI.
In other words, we cannot judge AI only by:
• Capability
• Speed
• Productivity
• Valuation
• Engagement
We must also judge it by what it does to:
• Human agency
• Work and livelihoods
• Truth and social trust
• Children and the vulnerable
• Inequality and exclusion
• Peace and the risk of war
"AI imitates intelligence, but it is not human"
This is a distinction the tech world often avoids because it feels philosophically inconvenient.
The Pope insists it matters.
AI does not have:
• A body
• Conscience
• Moral responsibility
• Lived experience
• Love, friendship, or inner growth
That's not an insult to the technology. It's a warning to us.
Because the moment we treat machines as moral agents, humans start escaping accountability.
A company can say: "the model decided."
A platform can say: "the feed optimized itself."
A military can say: "the target was selected algorithmically."
But none of those statements is morally sufficient.
Systems do not absolve people. Algorithms do not dissolve responsibility.
The real issue is power concentration
AI is not emerging in a neutral field.
The infrastructure—compute, chips, cloud, platforms, distribution, data—sits in very few hands.
When power concentrates, the risks are predictable:
• Opacity (nobody can explain what's happening)
• Dependency (societies become reliant on private systems)
• Manipulation (attention and belief can be engineered)
• Exclusion (some people become invisible)
• Inequality (benefits accrue to a small group)
This is where the Vatican's language and the investor's language meet.
Different words. Same reality.
Intelligence without accountability becomes power without restraint.
AI is not "immaterial magic"—it rests on real labor and real resources
One of the most grounded parts of the Pope's message is about work.
Behind every "instant" AI output is a physical and human supply chain:
• Data centers and energy
• Minerals and hardware supply chains
• Data labelers and trainers
• Content moderators
• Low-paid, invisible labor that makes the model look "smart"
The machine looks frictionless because the friction is pushed onto people we don't see.
Founders, investors, policymakers: this matters. The AI economy is not just software. It is labor, energy, hardware, and politics.
Truth is a civic issue, not a product issue
AI can generate content faster than societies can verify it.
That's not just "a misinformation problem."
It's a stress test for democracy and social trust.
When synthetic certainty floods public life, the boundary between truth and manipulation blurs—and people lose the shared ground needed to make collective decisions.
The hardest red line: autonomous weapons
The Pope is strongest here:
Lethal or irreversible decisions must not be delegated to opaque automated systems.
Once speed becomes the highest military virtue, human judgment becomes "the bottleneck" to remove.
That path is dangerous.
The deeper message is not anti-technology. It is anti-idolatry.
This is the part worth sitting with.
The encyclical does not reject AI.
It rejects the worship of AI.
It rejects the idea that:
• Progress is whatever increases capability
• Efficiency is the highest moral good
• Humans must justify their worth through productivity
• Private power can govern civilizational infrastructure by itself
So what do we do with this?
AI cannot be left only to the people racing to build it.
Again: not because they are bad people, but because no group should be builder, beneficiary, and regulator of a technology that reshapes work, truth, war, and human self-understanding.
We need outsiders—with real power and legitimacy:
• Scientists who can audit systems
• Regulators who understand incentives
• Workers who understand displacement
• Teachers who understand children's attention and development
• Philosophers and theologians who defend human dignity
• Investors who admit capital shapes behavior
• Citizens who refuse to outsource moral judgment
The right question is no longer "Can we build it?"
The right question is: What kind of human world are we building with it?
Here's your first step (next 24 hours)
Pick one area where you have leverage—company policy, investing criteria, governance, education, or media—and write down:
1. What is the human value we refuse to trade away?
2. What is the minimum oversight or constraint we will insist on?
3. Who are the "outsiders" we will invite in (and listen to) before we scale?
Clarity first. Speed second.
B Hari
Simplicity with substance
www.bhari.com