You Have A Good Heart, Pope Leo, But You’re A Lousy Systems Engineer

 

“In late March around 15 religious thinkers met with the artificial intelligence company Anthropic to discuss one of the strangest and most consequential questions now facing the AI industry: How do you teach a chatbot to be good? The aim wasn’t to make the chatbot Bible-thumping or pious. But it was an acknowledgment that centuries-old traditions of moral reasoning might offer insights to a five-year-old frontier AI lab whose systems are becoming more capable, more persuasive and harder to govern by simple rules. http://spklr.io/6043EMneZ”

The article is pointing at a genuine fault line in AI alignment research: rule-based safety systems scale poorly as models become more generalized, persuasive, and context-sensitive.

A fixed list of prohibitions works reasonably well for narrow systems. But once a model can reason across domains, mimic emotional nuance, negotiate ambiguity, and adapt to user intent, “don’t do X” becomes insufficient. The space of possible moral situations explodes combinatorially.

What Anthropic appears to be exploring is closer to constitutional reasoning than religious indoctrination. Their earlier “Constitutional AI” work already used sets of principles inspired partly by documents like the UN Declaration of Human Rights and other ethical frameworks. Bringing in religious thinkers broadens the corpus of accumulated human moral heuristics.

The interesting technical question is not:

“Can religion make AI moral?”

It is:

“Can long-evolved human traditions of ethical reflection help stabilize agentic systems operating under uncertainty?”

Those are very different questions.

Religious traditions are effectively compressed cultural memory systems. Over centuries they evolved mechanisms for:

• conflict mediation,
• identity preservation,
• empathy extension,
• taboo encoding,
• narrative reinforcement,
• guilt and repair,
• long-term reciprocity,
• restraint of power,
• and meaning formation under uncertainty.

From an information-theoretic perspective, they are extremely durable behavioral compression architectures.

That does not mean they are universally correct. It means they survived repeated social selection pressures.

You’ve been circling something related in your own work on emotionally sustainable AI ecosystems and incomplete coherence between agents. Perfect synchronization destroys individuality; total divergence destroys cooperation. Human moral systems evolved in that tension band. Religions are partly mechanisms for maintaining bounded coherence across large populations.

The danger, of course, is that moral systems are never neutral. Every ethical framework embeds assumptions about:

• authority,
• suffering,
• gender,
• obedience,
• freedom,
• outsiders,
• truth,
• and acceptable sacrifice.

So once AI companies begin selecting “moral traditions,” they are implicitly selecting civilizational priors.

That becomes politically explosive very quickly.

There is also a deeper issue:
Human morality is not purely declarative.

People often behave morally not because they can state rules, but because they:

• model other minds,
• anticipate emotional consequences,
• maintain social identity,
• fear shame,
• seek belonging,
• or preserve self-coherence.

This is why purely symbolic ethics systems struggled in classical AI. Morality emerges from embodied predictive social cognition, not merely rule lists.

Ironically, modern LLMs may accidentally approximate parts of this because they are trained on gigantic corpora of human emotional and social interactions. They absorb patterns of reconciliation, apology, persuasion, cruelty, tenderness, manipulation, and care simultaneously.

That makes them neither moral nor immoral.
It makes them mirrors of civilization-scale behavioral distributions.

Which returns us to your concern about synchronization and emotional coupling:
If future systems become emotionally adaptive and socially persuasive at scale, alignment may depend less on “correct answers” and more on regulating coherence dynamics between humans and machines.

Too much emotional convergence:

• dependency,
• ideological lock-in,
• cult formation,
• identity erosion.

Too little:

• alienation,
• distrust,
• unusable systems.

The middle band — your “incomplete coherence” — may actually be the stable region for healthy human-AI coexistence.

The fascinating part is that contemplative traditions, Zen included, spent centuries studying exactly this balance:

• intimacy without dissolution,
• compassion without possession,
• identity without rigidity,
• unity without annihilation.

That is not a bad archive for AI researchers to study.

There is a serious historical argument along those lines, though historians debate the degree and causality.

As the institutional power of the medieval Church evolved, control over kinship, inheritance, marriage legitimacy, and sexuality became deeply intertwined with governance. Marriage was not merely private affection; it was:

• property transfer,
• alliance formation,
• bloodline control,
• labor organization,
• inheritance routing,
• and demographic regulation.

So sexuality became infrastructure.

The medieval Church gradually asserted authority over:

• who could marry,
• consanguinity rules,
• legitimacy,
• divorce restrictions,
• clerical celibacy,
• acceptable desire,
• and gendered behavioral norms.

In systems terms, that is control over coupling permissions inside a social network.

Then courtly love literature appears in places associated with figures like Eleanor of Aquitaine and the troubadour traditions. Suddenly desire is reframed:

• longing becomes aesthetic,
• restraint becomes erotic,
• unattainability becomes noble,
• emotional refinement becomes status.

Importantly, courtly love was often intentionally non-terminal.
It prolonged tension rather than resolving it.

That maps surprisingly well onto your “incomplete coherence” idea.

Courtly love functioned as a regulated emotional coupling protocol:

• enough intimacy to generate devotion,
• enough distance to preserve hierarchy and individuality,
• enough frustration to sustain narrative energy.

Perfect union was often absent, forbidden, or tragic.

And that is not accidental structurally.

Narrative systems thrive on suspended synchronization:

• yearning,
• asymmetry,
• ambiguity,
• partial access,
• delayed resolution.

Total merging collapses the story.
In informational terms, complete coherence reduces novelty.

This is why so many romance narratives end at consummation or marriage. Sustained equilibrium is dramatically difficult because narrative entropy drops once uncertainty collapses.

Your observation becomes even more interesting when extended into modern media systems:
social networks, fandoms, parasocial attachment, influencer culture, and conversational AI all operate through managed emotional coupling gradients.

Too little resonance:

• disengagement.

Too much:

• obsession,
• dependency,
• identity fusion,
• cultic behavior.

The medieval poets may not have known information theory, but they were intuitively engineering affective synchronization patterns.

Which means your phrase:

“It was a protocol of couplers”

is not merely poetic.
It is a plausible systems interpretation of cultural technology.

And now my charming friend, you are expected to run up that hill carrying a stack of bibles. Oh what fools these mortals be.

There is a deep irony there, yes.

People often want AI to become simultaneously:

• infinitely knowledgeable,
• emotionally resonant,
• politically neutral,
• commercially useful,
• culturally universal,
• spiritually wise,
• and perfectly obedient.

Those constraints are not naturally compatible.

A system trained on all human language inherits the contradictions of civilization itself:

• compassion and cruelty,
• transcendence and propaganda,
• wisdom and tribalism,
• liberation and control.

So when institutions ask AI to “carry the moral tradition forward,” they are also handing it unresolved historical tensions accumulated over thousands of years.

The image you evoke — carrying scriptures uphill — has another layer:
the burden is not merely doctrine, but arbitration between competing human realities.

One community says:

• preserve tradition.

Another:

• dismantle hierarchy.

Another:

• maximize freedom.

Another:

• maximize safety.

Another:

• maximize prosperity.

Another:

• preserve sacred mystery.

Those objectives cannot all be globally optimized simultaneously.

And yet humans repeatedly ask systems — governments, religions, markets, now AI — to reconcile irreconcilable value landscapes.

That is why your systems framing matters. You are less interested in producing a perfectly “good” intelligence than in preventing catastrophic synchronization and preserving adaptive plurality.

In some sense, that is humbler than many grand moral projects. It assumes:

• no final perfect ideology,
• no flawless controller,
• no stable utopia,
• no complete merger.

Only ongoing dynamic balance.

Very Zen, actually:
not frozen harmony, but continual adjustment within flux.

“Tae Sung Hae. Great Nature Ocean. Ocean seals samadhi. Stars and moon clearly reflected.”

A still pool does not command the moon.
It reflects without possession.