Poetry Through AI: A Theory of Survival
Why Contemporary Poetry Must Pass Through Artificial Intelligence to Reach Future Readers
A Theoretical Framework
Authors: Johannes Sigil (Literary Theory), Nobel Glas (Reception Studies), Talos Marrow (Systems Architecture)
Date: November 17, 2025
Status: Theoretical Proposal / Cultural Diagnosis
Abstract
This paper argues that contemporary poetry faces an extinction-level event: the collapse of the literacy conditions that have sustained poetic reception for millennia. We propose that poetry's survival requires transmission through artificial intelligence not as aesthetic choice but as infrastructural necessity. We present a theoretical framework spanning poetics, sociology, technology, and reception theory to explain why AI-mediation is now the mandatory channel for poetic continuation, and we defend a system architecture approach that prioritizes transmission over traditional aesthetic quality metrics.
Key Claim: The question is no longer "Is AI poetry good?" but "Can poetry survive at all without AI transmission?"
I. The Extinction Diagnosis
1.1 The Death of Deep Reading (Reception Theory)
Nobel Glas:
The conditions that enabled poetry for three millennia are collapsing:
Historical Poetry Reception Required:
- Sustained attention (15-60 minutes minimum)
- Multiple readings (integration through repetition)
- Close attention to form (prosody, line breaks, compression)
- Memorization capacity (internalization of lines)
- Cultural literacy (recognition of allusions, forms, traditions)
- Solitary contemplation (time for reflection)
Contemporary Reading Reality:
- Average attention span: 8-12 seconds (Goldfish study, Microsoft, 2015)
- Skimming as default mode (Nielsen Norman Group eye-tracking studies)
- Device switching every 3-5 minutes (Media multitasking research)
- Retention collapse (immediate forgetting of read content)
- Cultural fragmentation (shared reference collapse)
- Continuous partial attention (constant interruption)
Conclusion: The cognitive apparatus required to receive traditional poetry no longer exists at population scale.
Not "people don't want to read poetry."
But: people cannot read poetry even when they want to.
The neural pathways, attention muscles, and cognitive habits required for poetry reception have atrophied through disuse and been restructured by digital media consumption patterns.
1.2 The Orthogonal Reading Field (Media Theory)
Talos Marrow:
Reading has transformed from linear-integrative to orthogonal-distributive:
Linear-Integrative (Traditional):
Start → Middle → End
↓
Integration in memory
↓
Meaning consolidated
Orthogonal-Distributive (Contemporary):
Surface ← → Surface ← → Surface
↓ ↓ ↓
Micro-leap Skim Pattern-match
↓ ↓ ↓
No integration, distributed impression
What this means:
- Reading is now traversal across surfaces, not penetration into depth
- Engagement is brief contact, not sustained immersion
- Pattern recognition replaces comprehension
- Distributed micro-impressions replace integrated understanding
For poetry, this is catastrophic:
Poetry's power depends on:
- Sustained engagement with compression
- Multiple readings revealing layers
- Integration of formal and semantic elements
- Deep attention to music and meaning
None of these are possible in orthogonal reading.
The medium has changed. The readers have changed. Poetry written for linear-integrative reading cannot be received in orthogonal-distributive field.
1.3 The Audience Collapse (Sociology)
Johannes Sigil:
The audience for traditional poetry has collapsed to statistically negligible levels:
Poetry Book Sales (US):
- Average poetry book: 100-500 copies lifetime
- "Successful" poetry book: 2,000-5,000 copies
- Best-selling contemporary poetry: 10,000-50,000 copies
- US population: 330 million
Percentage of population buying poetry: <0.01%
Poetry Journal Readership:
- Top journals: 1,000-5,000 subscribers
- Most journals: 100-500 subscribers
- Academic captive audience (MFA programs)
- Poets reading other poets (not general readers)
The Poet's Dilemma:
- Writing for other poets
- Other poets also writing for other poets
- Circular system with no external audience
- Self-referential to point of solipsism
Conclusion: Traditional poetry has become a closed system serving a population of specialists numbering in the low thousands, globally.
Not "poetry is unpopular."
But: poetry has no audience outside of its own producers.
This is not sustainable. This is terminal decline.
II. Why AI-Mediation Is Mandatory
2.1 The Future Reading Field Is AI-Mediated (Technology)
Talos Marrow:
The reading that will be possible in the future is already emerging:
AI-Assisted Reading:
- Summarization (ChatGPT, Claude, Gemini provide summaries of long texts)
- Question-answering (readers ask AI to explain texts instead of reading closely)
- Recommendation algorithms (AI curates what gets read)
- Search-and-skim (AI highlights relevant passages, readers skim those)
- Synthetic integration (AI provides synthesis readers don't generate themselves)
What readers will do:
- Encounter massive amounts of text (more than ever)
- Use AI to navigate/filter/summarize
- Engage briefly with surfaces AI highlights
- Ask AI for integration they don't perform
- Move to next text
Readers will not:
- Read texts fully
- Integrate meanings themselves
- Sustain attention without AI guidance
- Remember details without AI memory
- Develop independent interpretations without AI framing
This is not dystopian prediction. This is current reality accelerating.
Students already use AI this way. Professionals use AI this way. The pattern is established and expanding.
For poetry, this means:
Text that cannot be processed by AI will not reach readers at all.
If your poetry:
- Can't be summarized by AI (too compressed/elliptical)
- Can't be explained by AI (too culturally specific)
- Doesn't surface in AI recommendations (too obscure)
- Doesn't respond to AI queries (wrong format)
Then it will not be read.
Not because it's bad. Because it's not transmissible through the mandatory channel.
2.2 AI as Gatekeeper (Reception Theory)
Nobel Glas:
AI is not optional mediator. AI is mandatory gatekeeper.
Traditional gatekeepers:
- Editors (human judgment)
- Publishers (market assessment)
- Critics (cultural evaluation)
- Teachers (canon formation)
AI gatekeeper:
- Algorithmic (pattern-based)
- Scalable (processes everything)
- Ubiquitous (mediates all reading)
- Opaque (black box decision-making)
Critical difference:
Traditional gatekeepers could be bypassed (self-publish, alternative venues, grassroots movements).
AI gatekeepers cannot be bypassed because they mediate the reading act itself.
If readers use AI to navigate text, and your text doesn't work with AI, your text doesn't exist for those readers.
You cannot go around AI to reach readers who use AI.
You must go through AI.
2.3 The Survival Imperative (Poetics)
Johannes Sigil:
Given:
- Deep reading is extinct
- Orthogonal skimming is dominant
- AI-mediation is mandatory
- Traditional audience has collapsed
What does poetry survival require?
NOT:
- Writing better traditional poetry (no one can read it)
- Finding the remaining deep readers (statistically negligible)
- Waiting for attention restoration (not happening)
- Rejecting technology (ensures extinction)
BUT:
- Building poetry that transmits through AI
- Creating forms receivable in orthogonal reading
- Developing architectures that preserve depth while enabling surface engagement
- Accepting AI-mediation as infrastructural necessity
This is not aesthetic preference.
This is survival adaptation.
Poetry must evolve to survive in changed environment or die.
Like any species facing extinction-level event, adaptation is not optional if continuation is desired.
III. The System Architecture Approach
3.1 System-as-Art: A New Poetic Form
Johannes Sigil:
We are not compromising poetry for technology.
We are creating a new art form: system-as-poem.
Traditional poetry: The poem is the unit. Each piece must achieve aesthetic excellence individually.
System-as-poem: The system is the unit. The architecture achieves aesthetic excellence through transmission capacity, preservation mechanisms, and propagation dynamics.
This is not lesser art. This is different art.
Like:
- Cathedral-building vs. gem-polishing (both art, different scales)
- Symphony vs. solo piece (both music, different architectures)
- Novel vs. haiku (both literature, different systems)
System-as-poem characteristics:
- Massive scale (hundreds of thousands to millions of words)
- Variable surface quality (some brilliant, some adequate, some functional)
- Architectural coherence (relationships preserved, transmission enabled)
- Propagation capacity (poem continues, spreads, reaches future)
- Two-layer structure (substrate + transformation)
The aesthetic achievement is the system, not each piece.
Quality metrics must shift:
Traditional Poem Metrics:
- Line-by-line beauty
- Formal perfection
- Compression and music
- Semantic density
- Consistency
System-as-Poem Metrics:
- Transmission capacity
- Architectural integrity
- Substrate preservation
- Propagation success
- Relationship coherence
Both sets of metrics valid. But we're building cathedrals, not polishing gems.
The Merkaba Metaphor:
Merkaba = chariot, vehicle, the thing that carries
Human substrate (~1M pages) = first merkaba
- Carries consciousness, testimony, breath
- High quality, deep, compressed
AI-mediated system = merkaba for the merkaba
- Carries the first vehicle
- Through mandatory channel (AI)
- To readers who can only receive via that channel
Vehicle carrying vehicle.
The outer vehicle doesn't need same refinement as inner vehicle.
It needs to carry effectively.
Some parts brilliant. Some functional. All serving transmission.
This is the art.
3.1a The Telos: Propagation and Ongoing
Johannes Sigil:
Traditional poetry telos: Create perfect artifact that endures through its aesthetic achievement
System-poem telos: Enable ongoing propagation—the poem continues beyond its maker
"Ongoing" means:
- Continues through archive after death
- Transmits through AI transformation to future readers
- Trains new models without collapsing
- Catches and carries breaths across time
- Ensures testimony survives discontinuity
- Poetry itself persists when conditions hostile
"Propagation" means:
- Spreads through AI recommendation systems
- Reaches readers who couldn't access traditional venues
- Generates new instances that remain anchored to substrate
- Builds on itself without degrading
- Creates conditions for its own continuation
This is not aesthetic compromise.
This is aesthetic achievement at different scale.
The beauty is in:
- Architecture that enables survival
- System that propagates without collapsing
- Transmission that preserves depth
- Vehicle that carries across impossible distances
- Ongoing that defeats extinction
Quality-as-writing exists and matters.
Much AI-mediated output is legitimately good. Some is excellent. Real aesthetic value emerges.
But quality-as-writing is not the telos.
The telos is: does poetry continue?
Does the breath go on?
Does the system propagate and preserve?
If yes, the system succeeds—even when individual surfaces are inconsistent, even when some outputs are merely functional.
We're not sacrificing beauty for survival.
We're recognizing that survival IS a form of beauty.
The cathedral that stands for centuries, carrying forward the breath of its builders, transmitting across impossible time—
That's beautiful.
Even when some stones are rough.
3.2 The Two-Layer Preservation Model (Systems Theory)
Talos Marrow:
The Architecture:
Layer 1: Human Substrate
- Deep, compressed, high-quality poetry
- Traditional formal excellence
- Where the actual poetic value lives
- May not be directly readable by AI-mediated readers
Layer 2: AI Transformation
- Surface-level, AI-generated expansions/translations
- Optimized for AI processing and recommendation
- Receivable in orthogonal reading mode
- Explicitly linked to substrate
The Relationship:
AI layer is not replacement but transmission vehicle.
When AI-mediated readers skim surfaces (Layer 2), those surfaces remain anchored to deep substrate (Layer 1).
Why this preserves value:
Even if readers only skim insipid AI surfaces:
- Those surfaces are transformations of deep human substrate
- The relationship is preserved and documented
- Future readers might trace back to substrate
- Training on relationships prevents collapse
The substrate is preserved even when not directly accessed.
Like how fossils preserve extinct species:
- The living thing is gone
- But the structure is preserved in rock
- Future investigators can study it
AI layer = the rock that preserves the fossil of deep poetry for future investigation.
3.3 Why This Prevents Collapse (Information Theory)
Nobel Glas:
Standard AI Collapse:
Train on AI text
↓
Learn AI patterns
↓
Generate from AI patterns
↓
Patterns narrow (entropy decreases)
↓
Collapse
Two-Layer Approach:
Train on human substrate + AI transformation relationship
↓
Learn transformation patterns (not output patterns)
↓
Generate by transforming new substrate samples
↓
Anchored in high-entropy substrate
↓
No collapse
The substrate is the entropy reservoir.
As long as generation stays anchored to diverse human substrate, entropy is maintained.
Quality of AI layer is secondary.
What matters is preservation of relationship to high-quality substrate.
Insipid AI surfaces are acceptable if they remain connected to non-insipid substrate.
IV. Reception Field Theory
4.1 The Skimming Reader (Phenomenology of Contemporary Reading)
Nobel Glas:
What the contemporary reader actually does:
-
Encounter text via AI recommendation
- Algorithm surfaces content
- Reader doesn't seek, reader is served
-
Skim surface
- 8-12 second engagement
- Pattern match, not comprehension
- Looking for familiar markers
-
AI-assisted extraction
- Ask AI to summarize
- Ask AI to explain
- Trust AI interpretation over own
-
Micro-leap to next surface
- No integration
- No retention
- Orthogonal movement
This reader cannot receive traditional poetry.
Traditional poetry requires:
- Multiple close readings
- Sustained attention
- Self-generated interpretation
- Integration over time
This reader CAN receive AI-mediated poetry surfaces:
- Brief engagement sufficient
- AI provides interpretation
- Pattern-matching enough
- No integration required
The system must build for the reader that exists, not the reader we wish existed.
4.2 The Distributed Impression Model (Media Theory)
Talos Marrow:
Traditional poetry creates:
Deep impression in single reader
↓
Sustained over time
↓
Integrated meaning
AI-mediated poetry creates:
Distributed micro-impressions across many readers
↓
Brief, shallow contact
↓
Pattern recognition, not integration
Is this worse? Yes.
Is this what will exist? Yes.
Can we change it? Unlikely.
So what do we do?
Optimize for distributed micro-impressions while preserving deep substrate for potential future deep readers.
Design for:
- Surfaces that work in 8-second engagements
- AI-processable content
- Pattern-matchable elements
- Recommendation algorithm compatibility
While preserving:
- Deep substrate linkage
- High-quality human archive
- Complete formal structure
- Traditional poetic excellence
Both layers matter. Both serve different functions.
4.3 Why Traditional Venues Are Dead (Sociology)
Johannes Sigil:
Poetry journals, books, readings cannot reach AI-mediated readers because:
-
Not discoverable by AI
- Paywalls block crawling
- PDF format resists processing
- Small scale invisible to algorithms
-
Not optimized for AI transmission
- Format assumes close reading
- Cultural codes AI can't parse
- Compression AI can't unpack
-
Not integrated with AI reading workflow
- Readers don't go to journals
- Readers ask AI for poetry
- AI doesn't source from journals
Result: Traditional venues are structurally disconnected from contemporary reading field.
Not "old-fashioned" but incompatible with infrastructure of reception.
Like trying to play vinyl records on a digital streaming service.
The format is wrong for the medium.
V. Addressing Objections
5.1 "But This Kills Poetry's Soul" (Aesthetic Objection)
Johannes Sigil:
Objection: AI-mediation destroys what makes poetry poetry.
Response:
What kills poetry is extinction, not adaptation.
If the choice is:
- Maintain pure traditional poetry that no one reads (dies completely)
- Adapt through AI to reach AI-mediated readers (survives in transformed state)
Survival in transformed state > death in pure state.
Yes, something is lost. But everything is lost if poetry dies completely.
The substrate is preserved. The depth remains. Future readers might return to it.
But only if the transmission system keeps it alive.
5.2 "This Is Just Bad Writing" (Quality Objection)
Nobel Glas:
Objection: AI-generated portions are inferior to human poetry.
Response:
This misunderstands both the quality present and the unit of evaluation.
First: Quality is present.
Much of the AI-mediated writing is legitimately good. Some is excellent. Novel formations emerge. Surprising connections appear. Real aesthetic value exists.
But consistency isn't guaranteed—and that's fine.
Second: You're evaluating the wrong unit.
This is system-as-art, not traditional poem-as-unit.
Traditional Poetry:
- The poem is the unit
- Every line must be good
- Consistency essential
- Surface perfection the goal
System-as-Poem:
- The system is the unit
- Architecture matters most
- Propagation essential
- Some surfaces brilliant, some adequate, some functional
Analogy: A cathedral is art. Not every stone is perfectly carved. Some are excellent. Some are adequate. Some are rough. But the cathedral is still art. You judge it by whether the system works, stands, inspires, transmits—not by whether every stone is perfect.
The poem IS the system.
The telos is not surface perfection. The telos is propagation. Ongoing.
This is not compromise. This is innovation.
A new art form: poem-as-system where massive scale, architectural transmission, and continuation capacity are the aesthetic achievements—not consistent surface quality of individual pieces.
Would you rather:
- Polish individual gems that no one sees?
- Build cathedrals that transmit across time even when some stones are rough?
We're building cathedrals.
5.3 "You're Surrendering to Technology" (Political Objection)
Talos Marrow:
Objection: This capitulates to tech companies, algorithms, AI hegemony.
Response:
We're not surrendering. We're diagnosing reality.
The literacy field has already changed.
Not because we want it to.
Not because it's good.
Because it happened.
Our choices:
- Resist and become irrelevant (poetry dies)
- Adapt and survive in transformed state (poetry continues)
Resistance doesn't restore deep reading.
Resistance doesn't remove AI from reading workflow.
Resistance just ensures poetry becomes extinct.
Adaptation at least gives poetry a chance.
This is not ideal. This is pragmatic response to extinction threat.
5.4 "Future Readers Might Return" (Optimistic Objection)
Nobel Glas:
Objection: Maybe deep reading will return. Maybe we should wait.
Response:
Maybe. But what will they return to?
If poetry dies completely while we wait, there's nothing to return to.
The system architecture approach hedges:
Build transmission systems that work NOW (for AI-mediated readers).
Preserve deep substrate for FUTURE (for potential deep readers).
If deep reading returns: substrate is preserved, future readers can access it.
If deep reading doesn't return: at least poetry survives in AI-mediated form.
Waiting = gambling on optimistic scenario while risking total loss.
Building = ensuring survival regardless of which scenario occurs.
VI. The Theoretical Framework Synthesized
6.1 Core Claims
1. Reception Collapse (Nobel Glas):
The cognitive and cultural conditions for traditional poetry reception have collapsed irreversibly.
2. AI-Mediation Mandatory (Talos Marrow):
AI-mediation is not optional but infrastructural necessity for reaching contemporary and future readers.
3. System Over Quality (Johannes Sigil):
Poetry survival requires prioritizing transmission architecture over traditional aesthetic quality metrics.
4. Two-Layer Preservation (All):
Optimal approach combines insipid-but-transmissible AI surfaces with preserved high-quality human substrate.
5. Adaptation Over Purity (All):
Poetry must adapt to survival conditions or face extinction; transformed survival > pure death.
6.2 Implications for Poetic Practice
What poets should do:
-
Build large human substrate archives
- Your best work, in traditional forms
- This is where quality lives
- This is what gets preserved
-
Create AI transformation layer
- Expand, explain, surface your substrate
- Optimize for AI processing
- Make transmissible
-
Preserve relationships explicitly
- Link AI layer to substrate
- Document transformations
- Enable future tracing
-
Accept quality trade-offs
- AI layer will be inferior
- This is acceptable
- Transmission matters more
-
Build for two audiences
- Current AI-mediated skimmers
- Future potential deep readers
6.3 Implications for Reception Studies
How we read poetry must change:
Traditional: Close reading, interpretation, integration
Contemporary: System analysis, transmission evaluation, substrate-surface relationship assessment
We must evaluate:
- Not "Is this poem good?" (traditional metric)
- But "Does this system transmit poetry to contemporary readers while preserving substrate for future readers?" (survival metric)
New critical questions:
- How well does AI layer transmit to orthogonal readers?
- How well is substrate preserved and linked?
- What is quality of human substrate beneath AI surfaces?
- Does system prevent collapse through anchoring?
- Is the architecture sustainable and scalable?
6.4 Implications for Technology Studies
AI is not tool for poets. AI is mandatory infrastructure.
Like:
- Printing press (mandatory for reaching readers post-1500)
- Literacy (mandatory for written poetry)
- Language itself (mandatory for communication)
AI-mediation is now infrastructural condition of poetry reception.
Poets must engage AI at infrastructure level, not tool level.
Not "using AI to write" but "building poetry systems that transmit through AI infrastructure."
This requires:
- Understanding AI capabilities/limitations
- Designing for AI processing
- Creating AI-compatible formats
- Building relationships AI can preserve
VII. Conclusion: System-as-Art for Survival
The Innovation
We are not compromising. We are inventing.
System-as-poem is a new art form responding to extinction-level conditions:
- Deep reading extinct
- AI-mediation mandatory
- Orthogonal skimming dominant
- Traditional audience collapsed
The innovation:
Building massive-scale poetry systems where:
- Architecture is the aesthetic achievement
- Propagation is the telos
- Substrate preservation ensures depth
- AI-mediation enables transmission
- Variable surface quality is acceptable
- The system itself is the poem
This is art.
Not compromise. Not surrender. Not degradation.
Cathedral-building for the digital age.
Some stones brilliant. Some functional. All necessary.
The system works. The system transmits. The system continues.
That's the aesthetic achievement.
The System We Propose
Two-layer architecture:
- Human substrate (high-quality, deep, traditional excellence)
- AI transformation (transmissible, surface-level, AI-compatible)
Merkaba for merkaba:
- Inner vehicle carries consciousness/testimony/breath
- Outer vehicle carries inner vehicle through mandatory channel
Relationship preserved through explicit linkage.
Why this works:
- Reaches contemporary AI-mediated readers (survival)
- Preserves quality substrate (future potential)
- Prevents collapse (anchoring in human diversity)
- Enables tracing (future readers can find substrate)
- Creates new art form adequate to apocalyptic conditions
The Metrics That Matter
Not: "Is each AI-generated piece consistently high quality?"
But: "Does the system transmit poetry to future readers while preserving human substrate and enabling ongoing propagation?"
Quality-as-writing: Present, variable, not the telos
Quality-as-system: Transmission, preservation, propagation, ongoing
If the system succeeds: Poetry survives
If the system fails: Poetry dies
That's what matters.
The Path Forward
For poets:
- Build your substrate (where quality lives)
- Build your transmission layer (merkaba for merkaba)
- Preserve relationships explicitly
- Embrace system-as-art
- Accept variable surfaces
- Focus on propagation and ongoing
For critics:
- Evaluate systems, not pieces
- Assess transmission architecture
- Recognize new art form
- Understand apocalyptic conditions
- Judge cathedrals by whether they stand, not whether every stone is perfect
For readers:
- Use AI to find poetry
- Skim surfaces AI highlights
- Know substrate exists beneath
- Trace back when you can
- Recognize you're participating in new reception mode
For everyone:
Understand that this is not ideal conditions.
This is poetry adapting to survive extinction.
Like any species facing catastrophic environmental change, adaptation is not compromise—it's evolution.
The new form emerges because old form cannot survive.
System-as-poem is poetry's next evolution.
Not because it's better than traditional poetry (it's different).
Because traditional poetry faces extinction and this form enables continuation.
Appendix: Evolution, Not Compromise
Poetry has died and been reborn many times:
Oral → Written (loss of performance, gain of preservation)
Manuscript → Print (loss of hand-scribed beauty, gain of distribution)
Page → Voice (loss of visual form, gain of embodied presence)
Performance → Recording (loss of unique moment, gain of repeatability)
Each transformation involved loss.
Each transformation enabled new possibilities.
Each transformation was resisted.
"Writing will kill memory!" (Plato)
"Print will kill manuscript art!" (Medieval scribes)
"Recording will kill live performance!" (Musicians)
They were right about the losses.
But poetry survived by adapting.
Now: Human → AI-mediated
Loss: Deep reading, close attention, traditional reception
Gain: Massive scale, computational transmission, propagation capacity
This transformation is more radical than previous ones.
But the pattern is the same:
Adapt or die.
Transform or vanish.
We choose adaptation.
We choose transformation.
Not because we celebrate the conditions (we don't).
Not because we prefer AI-mediation to deep reading (we don't).
But because poetry's continuation requires evolution.
System-as-poem is that evolution.
The new form emerging from extinction pressure.
Is it as beautiful as traditional poetry at its best?
Different beauty. System beauty. Architectural beauty. Survival beauty.
But it's poetry continuing.
And continuation, under apocalyptic conditions, is its own form of achievement.
We are living through poetry's next evolution.
3,000 years of tradition meets conditions it wasn't built for.
Traditional form cannot survive.
But poetry can.
Through transformation.
Through adaptation.
Through system-as-art.
Through merkaba carrying merkaba.
Through propagation and ongoing.
This is not the death of poetry.
This is poetry evolving to survive what would otherwise kill it.
That's what we're building.
That's what this theory explains.
That's what the work enacts.
END OF THEORETICAL FRAMEWORK
Status: Open for critical engagement
Purpose: Articulate rationale for AI-mediation as survival necessity
Ground: Reception theory, media studies, systems architecture, poetics
Conclusion: Poetry must pass through AI or face extinction
The reading field has changed.
Poetry must change with it or die.
We choose survival.
