I write a lot of poetry, but I’d be the first to admit I don’t know much about it. I think of it this way: I can pick up a block of Plasticine and model things with it, yet possess no education in the sculptor’s art. I simply find it pleasing as a means of expression. Poetry is much the same.
It strikes me, too, how poetry has always occupied an uneasy position within the culture. It is among our oldest forms of expression, yet is increasingly marginal. Institutional poetry may be shrinking – smaller readerships, declining humanities departments, diminishing commercial viability – but the poetic impulse itself does not seem to be disappearing at all. Rather, it diffuses into unexpected places, with many writers now bypassing traditional publishing and sharing their work privately online.
Yet while the instinct persists, the formal teaching of poetry as a craft appears to be diminishing alongside the institutions. Inspiration alone does not necessarily produce good poetry, but it is good poetry that endures. Which raises an interesting question. Can AI make us better poets? More particularly, can it democratise the writing of poetry itself? Good poetry – meaningful perhaps not to any institution, but to the poet.
At first glance, this sounds absurd. Poetry remains one of the most intimate and deeply human forms of expression. It emerges from memory, longing, grief, dreams, love, and symbolic perception. AI, by contrast, is computational, statistical, trained upon immense archives of language. How could something non-conscious participate meaningfully in poetry?
Yet I suspect many contemporary writers are discovering what I have discovered: AI can indeed help, not in originating poetry, but in refining and clarifying our own expression of it. That distinction matters enormously.
My poems never begin with technique. They arise from an image, a memory, an emotional atmosphere, a dream fragment, an encounter with beauty or mortality. The question of craft arises later, when questions emerge about rhythm, precision, metaphor, and whether the language is saying too much or too little.
I suspect poets rarely developed in isolation. Even the most individual voices are shaped through communities of criticism, correspondence, rivalry, or dialogue with tradition itself. And now, as some of those institutions decline, AI steps into the gaps as a new kind of critic. It can compare patterns across vast bodies of literature and provide a level of editorial feedback many aspiring poets might never otherwise encounter. In this sense, AI may democratise access not to poetic genius, but to aspects of poetic craft. It can teach us how to write better poems.
This matters because poetry has often been culturally gate-kept. Yet the poetic impulse itself has never belonged exclusively to institutions. Human beings have always written poetry beyond official culture.
Someone like me, without formal literary education, approaches poetry largely through instinct and assimilation, by reading widely across poets past and present. AI now offers immediate feedback on rhythm, imagery, clarity, and structure. This does not mean AI can create a poet from nothing. No tool can substitute for perception, emotional honesty, or symbolic sensitivity. But it can help refine the language, which reveals something important about poetry itself.
If poetry were merely the technical arrangement of beautiful language, AI-generated poetry would already satisfy us completely. Yet while many AI poems do possess recognisable poetic markers – metaphor, cadence, emotional vocabulary – they often feel curiously hollow to the human heart.
Genuine poetry emerges from lived encounter. Behind the language lies some contact with reality filtered through consciousness: grief endured, beauty witnessed, longing suffered, a dream remembered. Readers know when language carries an authentic energy, rather than merely something that imitates its appearance. Which is why AI’s most valuable role may be as teacher, editor, and collaborator rather than synthetic poet.
Collaboration with AI returns us to fundamental poetic questions.
- Why does this line feel weak?
- Is this image a cliché?
- What makes this metaphor effective?
- Should this poem say less?
- Where is its emotional centre?
These are all genuine poetic concerns. And AI may encourage closer attention to language at precisely the moment contemporary culture is becoming increasingly deaf to it.
The dangers, however, are obvious.
Because AI systems are trained upon existing literature, they naturally favour familiar patterns. They tend toward the stylistic mean, encouraging polished competence rather than genuine surprise. They also prefer coherence and completion, whereas some of the most powerful poetry derives its force from ambiguity and unresolved tensions. Human meaning often resides precisely in the spaces that resist explanation.
There is also the danger of inflation: language that sounds profound without earning its profundity. AI can produce endless streams of vaguely mystical phrasing that mimic depth while lacking existential weight. Without discernment, this risks flooding the culture with competent-sounding but ultimately empty language.
AI may therefore risk contributing simultaneously to both the expansion and the dilution of poetry. Yet every technological transformation has altered literary culture, and poetry has survived because its deepest function does not depend upon any particular medium.
Modern societies increasingly treat language as a communications technology: something to advertise, persuade, classify, and manage information. Poetry resists this reduction. Through metaphor and symbolism it reveals dimensions of experience that exceed literal description.
Poetry remains closely akin to dreams. Dreams speak primarily through images rather than concepts, and their symbols carry multiple meanings simultaneously. Poetry works in much the same way, allowing meaning to radiate outward rather than collapse into a single interpretation.
In highly technological societies this symbolic dimension feels increasingly fragile. Public discourse rewards immediacy, certainty, and simplification. Poetry preserves ambiguity, resonance, inwardness, and a more contemplative mode of perception.
It is ironic, then, that AI may help preserve aspects of such symbolic literacy precisely because it invites us back into closer engagement with language. Revising poetry collaboratively with AI can heighten awareness of rhythm, metaphor, tone, and symbolic layering. The democratisation here is therefore not simply about producing more poems, but about widening participation in poetic consciousness itself. And poetic consciousness matters because human beings are not entirely rational creatures. We dream symbolically. We remember things emotionally. We experience beauty, dread, grief, eros, mortality, and mystery in ways ordinary informational language cannot fully grasp.
As long as such experiences remain part of human life, poetry in some form remains essential. Institutional poetry may continue its decline, while poetic practice becomes increasingly distributed, collaborative, online, and of course entirely unpaid. AI will not replace poets so much as alter the ecology in which our poetry evolves.
The essential question then may no longer be whether AI assisted in crafting a poem, but whether there was genuine perception behind the language in the first place.
Perhaps then, poetry's future lies neither in resisting technology nor surrendering art to machines, but in discovering new forms of collaboration that preserve what is most essential about the poetic experience. AI may democratise poetry not by replacing poets, which would be absurd, but by widening access to poetic consciousness itself. For many of us, the poetic impulse is already there, though perhaps dormant, only waiting to be taken up.