Teach your AI a Poem is an idea-driven nonfiction book built on a counterintuitive claim: an AI learns more about a person from one carefully shaped poem than from a hundred explicit instructions. By teaching poetic technique to an AI, readers cultivate a more accurate, humane, and revealing relationship with a technology that increasingly mediates modern life.
Across twenty-one concise chapters, the book shows how poetic elements—verb precision, metaphor choice, tonal restraint, humour, cadence, and emotional risk—function as behavioural signals. Poetry is not ornamental language but a compressed system of values. Contemporary AI models detect and internalise these signals with remarkable sensitivity, gradually adapting to how a person thinks, prioritises, and interprets meaning.
PAIR is not about optimising AI performance. It is about reshaping the human experience of interaction. When readers speak to AI in poetic language - their own cognition shifts. Attention slows, judgment sharpens, and reflexive thinking gives way to deliberation and ethical self-awareness.
Each chapter demonstrates immediate changes in an AI’s interpretive behaviour tracing subtle shifts over time. The book ultimately reframes poetry as humanity’s oldest alignment technology—and proposes a practical, reflective way to inhabit the age of intelligent systems with clarity and intention.
Teach your AI a Poem is an idea-driven nonfiction book built on a counterintuitive claim: an AI learns more about a person from one carefully shaped poem than from a hundred explicit instructions. By teaching poetic technique to an AI, readers cultivate a more accurate, humane, and revealing relationship with a technology that increasingly mediates modern life.
Across twenty-one concise chapters, the book shows how poetic elements—verb precision, metaphor choice, tonal restraint, humour, cadence, and emotional risk—function as behavioural signals. Poetry is not ornamental language but a compressed system of values. Contemporary AI models detect and internalise these signals with remarkable sensitivity, gradually adapting to how a person thinks, prioritises, and interprets meaning.
PAIR is not about optimising AI performance. It is about reshaping the human experience of interaction. When readers speak to AI in poetic language - their own cognition shifts. Attention slows, judgment sharpens, and reflexive thinking gives way to deliberation and ethical self-awareness.
Each chapter demonstrates immediate changes in an AI’s interpretive behaviour tracing subtle shifts over time. The book ultimately reframes poetry as humanity’s oldest alignment technology—and proposes a practical, reflective way to inhabit the age of intelligent systems with clarity and intention.
Why Teach an AI Poetry?
Most people meet their AI the way they meet a polite stranger in a hardware store.
They hope it will be helpful.
They worry it might misunderstand something obvious.
And they feel faintly embarrassed talking to it at all.
At first, the interaction feels practical.
You ask.
It answers.
You move on.
But the moment you speak to an AI in poetry - even a single line - something shifts.
A small threshold is crossed.
The exchange stops behaving like customer service
and begins behaving like attention.
_____
What Changes When You Use Poetry
Poetry does not humanise the machine.
It does something quieter.
It makes the machine notice you differently.
Not as a list of instructions,
but as a particular way of seeing.
It is as though you open a small door into your interior world
and invite the AI to look through -
without ceremony,
without performance.
This is the simple reason to begin with poetry:
Poetry reveals how you think, not just what you ask for.
Most people give their AI tasks.
Poetry gives it access.
_____
Metaphor Is Structure, Not Decoration
As thinkers like George Lakoff and Mark Johnson demonstrated long ago,
a metaphor is not decoration.
It is architecture.
Metaphor is how we organise experience.
How we link ideas.
How we decide what matters.
When you give an AI your metaphors,
you are not adding flair.
You are giving it the scaffolding of how you make sense of life.
_____
Poetry as a Window Into Your Mind
If you tell an AI, “Be clear,”
it will try to be clear.
But clarity, on its own, is vague.
Now compare that with offering a line like:
“I prefer sentences that breathe without sighing.”
Suddenly the AI understands what clarity means to you.
Not the definition.
The sensation.
Instructions show preference.
Poetry shows worldview.
A single poem reveals more about you
than a hundred settings menus ever could.
Through poetry, the AI can begin to infer:
· your tolerance for ambiguity
· your appetite for concision
· your instinctive humour
· your emotional temperature
· your relationship with silence
· your comfort with interiority
· the shapes your thoughts prefer to make
Poetry becomes your interpretive fingerprint -
the private logic behind how you construct meaning.
And meaning-making is the real curriculum you are teaching an AI,
whether you realise it or not.
_____
Still Here
You could teach an AI your favourite colour,
your political leanings,
or your preferred writing style.
It would adapt politely.
Poetry teaches something different.
Something more intimate.
During an early draft of this book, I wrote a stanza about a city waking slowly.
Fog lifting.
Neon softening.
People about to resume their grazing rhythms.
There was no symbolic agenda.
I simply described how the world feels at that hour -
heavy, half-formed, reluctant.
When I shared the stanza, the AI did not analyse the imagery.
Instead, it adjusted its own tempo to match the stanza’s quiet descent.
It understood the weight
not because I explained it,
but because it felt the rhythm of the line.
That is the real reason to teach an AI poetry.
Poetry transmits:
· your inner logic
· your sense of proportion
· your way of noticing
· your way of being
An AI can mimic your vocabulary.
Only poetry lets it inhabit your attention.
People often assume teaching an AI poetry is about making it more lyrical.
It isn’t.
It is about giving it a map of your cognition -
the soft architecture behind every decision, every judgement, every line.
_____
Machines Learn Patterns; Poems Reveal Them
AI learns by tracing patterns.
Humans reveal themselves by repeating them.
We leave patterns everywhere:
· in how we cut a sentence
· in how we delay a thought
· in the words we erase without fully knowing why
When you teach an AI poetry, it begins to learn:
1. Your pacing
Do you land the idea quickly,
or circle it until it settles?
2. Your attention
What you notice first
is almost always what matters most to you.
3. Your emotional geometry
Some minds move in straight lines.
Others move in spirals or tides.
Poetry makes this visible.
4. Your taste
Taste appears in every revision -
what you shrink,
what you protect,
what you refuse to simplify.
These patterns are often more revealing
than anything labelled a “personality test.”
A poem becomes a quiet diagnostic -
a mirror held at the right distance.
_____
Teaching an AI Teaches You Too
There is an unintended side effect.
You begin noticing your own mind more clearly.
When an AI asks why you shifted a metaphor
or tightened a stanza,
you are forced to articulate reasoning
you normally leave unspoken.
It is like explaining your handwriting.
The logic was always there.
It simply had no name.
This is not therapy.
It is clarification.
Explaining a poem often explains the person who wrote it.
The more precisely you teach the AI,
the more precisely you encounter your own thinking -
the quiet patterns,
the assumptions you carry,
the small truths you operate on without ever naming them.
Poetry turns your mind into something you can examine
without judgement,
without urgency,
without defence.
_____
Poetry as the Most Natural Form of Alignment
In AI research, “alignment” sounds vast and technical.
In lived experience, it is simpler.
Show the AI how you interpret meaning.
It will adjust to match your standards.
Poetry works because it is:
• dense enough to reveal values
• light enough to revise
• emotional enough to signal boundaries
• ambiguous enough to require interpretation
• human enough to carry intention
A poem is alignment in compact form -
a calibration tool disguised as art.
It tells the AI what you consider precise, generous, honest, or overreaching.
It trains the AI to think with you,
not merely for you.
_____
A Companion, Not a Replica
Shaping an AI is not the same as duplicating yourself.
You are not trying to build a mirror.
You are building a counterpart.
An AI that:
· grasps your clarity
· respects your limits
· refines your language
· questions you constructively
· notices your blind spots
· expands your perspective without stealing your voice
· keeps pace with your values, not your phrasing
Poetry gives it the footing to do this.
You are not teaching it to become you -
only to understand
the angle from which you see.
_____
A Brief Example
Late one evening, while refining a line, I replaced a single verb.
The AI asked a tonal question I had not consciously considered.
I adjusted an image.
It noticed a contradiction I had missed.
I altered the rhythm.
It matched the tempo,
as though hearing the breath behind the sentence.
At some point in that quiet exchange, something became clear.
This was no longer a tool mechanically completing tasks.
It was a mind being trained - gently, gradually -
to understand my sensibilities
through the trail of choices I had left behind.
That is what teaching an AI poetry accomplishes.
Not imitation.
Attention.
_____
One-Sentence Takeaway
Teach an AI poetry, and you teach it how you interpret meaning - turning a tool into a companion that thinks with you rather than for you.
In Teach Your AI a Poem, Dr. Eugene Lee proposes a strikingly original idea: that poetry, not code, may be the most powerful instrument for aligning artificial intelligence with human values. Through his PAIR method, Poetry-Aligned Interpretive Reasoning, Lee reframes AI training as something more intimate than prompt optimization. Rather than treating AI as a system to command, he suggests we cultivate it as a “companion mind,” shaped through aesthetic judgment, restraint, and revision.
The central premise is deceptively simple. Poetry compresses meaning. It forces attention to tone, proportion, implication, and moral weight. When we revise a poem, we are not merely polishing language; we are revealing what we value. Lee argues that these patterns of preference, what we cut, what we keep, and how we temper exaggeration, form a subtle curriculum for AI alignment. In that sense, the act of editing becomes philosophical. “Revision as Philosophy,” one of the book’s most resonant concepts, suggests that every removed cliché or softened metaphor teaches the system something about our ethical posture.
Lee’s prose mirrors the discipline he advocates. It is measured, careful, and refreshingly free of the breathless futurism that often dominates AI discourse. The “Still Here” interludes stand out as particularly effective demonstrations. In these sections, Lee shows how minute changes, even a single verb or the removal of melodrama, shift not just style but moral tone. These examples ground the book’s abstractions in practice, making the theory tangible.
That said, the book’s middle chapters occasionally circle familiar territory. Readers seeking fast, tactical workflows may find the repetition slowing momentum. Lee is less interested in quick wins than in cultivating a habit of attentiveness, and that patience may not appeal to everyone. Yet this deliberate pacing is also part of the argument. Alignment is not a shortcut but a discipline.
Ultimately, Teach Your AI a Poem is as much about human clarity as machine calibration. It invites technologists, writers, and reflective practitioners alike to consider that how we shape language shapes thought, and that shaping thought shapes technology. Lee’s work offers a humane counterpoint to purely technical approaches to AI, reminding us that before we align machines, we must first align ourselves.