by Marlon Barrios Solano
William Carlos Williams (1883–1963) stands as a central figure of American modernist poetry, celebrated for his bold departures from inherited European forms. A practicing physician by day, Williams forged a distinctly American poetic voice in the evenings, one that renounced lofty diction and strict meters in favor of everyday language and free rhythms. He believed poetry should capture the American idiom — the plain, vivid speech of common people — and thereby reveal universal truths through concrete particulars. “The poet’s business,” Williams wrote, “is not to talk in vague categories but to write particularly, as a physician works, upon… the thing before him in the particular to discover the universal.” This insistence on the particular was Williams’s rebellion against what he called the “dead” forms and diction of the past.
Instead of imitating European masters or relying on traditional meter, he rooted his poems in immediate perception — capturing objects and moments of daily life in fresh terms (“no ideas but in things,” as he famously put it in his epic poem Paterson). Through this approach, Williams helped inaugurate an American modernism that was boldly experimental yet grounded in local reality. His short imagistic lyrics (like “The Red Wheelbarrow” or “This Is Just To Say”) and his sprawling epics (Spring and All, Paterson) alike exemplify a poetics of innovation: each new poem had to find its own form, driven by a new way of seeing.
Williams’s resistance to inherited forms — whether the sonnet’s “staid concatenations” or the rigid iambic pentameter — was not mere contrarianism, but a principled belief that new content demanded new form. In his 1948 lecture “The Poem as a Field of Action,” he argued that while modern life introduced new subjects into poetry, the measure and structure of English verse had not kept pace; to truly be modern, poetry must revolutionize its technique from the ground up. Williams thus pioneered techniques such as variable meter and the visual spacing of words, seeking what he called the “variable foot” — a flexible, native American cadence to replace the old iambic foot.
In all these innovations, one principle remained constant: a poem’s form should be an extension of the poet’s mind in action. This conviction is encapsulated in a deceptively simple maxim that haunts Williams’s work and legacy:
A new line is a new mind.
The phrase “A new line is a new mind” and its variants can be traced directly to Williams’s writings, emerging most notably in his long poem Paterson. In Paterson, Book II (published 1948), Williams makes a striking pronouncement:
“Unless there is a new mind there cannot be a new line, the old will go on repeating itself with recurring deadliness.”
This line appears in the context of Williams’s meditation on invention in poetry. He warns that without fresh thought, formal experimentation becomes impossible — the poet will merely rehash old rhythms and clichés, yielding lifeless verse. The Paterson quote crystallizes Williams’s credo that genuine innovation in poetry begins internally, in the realm of ideas and perception. A poet must cultivate “a new mind” — a new way of observing and knowing the world — in order to write a truly new line of poetry. Form and content are thus inextricably linked: “the old will go on repeating itself” unless one’s mode of thinking breaks free of inherited patterns.
Williams not only declared this principle in Paterson; he also dramatized it throughout that poem’s collage-like structure. Paterson itself is a text constantly breaking into new forms — shifting from lyric verse to prose letters to newspaper clippings — enacting the idea that each new section demands a renewed mind. At one moment, the poem’s narrator urges himself to push beyond conventional language:
“No ideas but in things”
— a call to find a new line by grounding thought in fresh perceptions of reality. The “new mind/new line” dictum thus serves as both a thematic statement within Paterson and a formal method driving its composition.
Variations of this idea surface elsewhere in Williams’s oeuvre, underscoring how fundamental it was to his poetics. In one of his late poems, “To Daphne and Virginia” (written for his two daughters-in-law and published in Pictures from Brueghel and Other Poems, 1962), Williams revisits the notion with a slight twist of phrasing:
“A new world is only a new mind.
And the mind and the poem are all apiece.”
Here the scope of the claim widens from line to world: a new world (a new reality or experience) comes into being only through a new mode of mind, and crucially, the poem itself shares in that same mind. The poem is the mind, in poetic form — “all apiece” with it. In Williams’s view, then, a genuinely new poem creates a new way of seeing the world, just as a genuinely new way of seeing creates a new poem. The line from “To Daphne and Virginia” condenses decades of Williams’s artistic practice into an aphorism: the boundaries of our world are the boundaries of our cognition; by extending one, we extend the other.
Notably, Williams frames this insight as an address to loved ones, lending it a personal, almost instructional tone. Be patient, he begins in that poem, that I address you in a poem… The mind lives there. It is uncertain, can trick us… But for resources what can equal it? He is, in effect, teaching that in poetry (as in life) the mind’s inventiveness is our greatest resource — it alone can break us out of old ruts and provide the “wings” to fly to new understandings. When he finally declares “A new world is only a new mind,” it comes as the hard-won lesson of the poem.
These linked variations — new mind → new line in Paterson, and new mind → new world in the later poem — reinforce one another. Together, they sketch Williams’s philosophy of creativity: true novelty in art or experience cannot be achieved by rearranging old forms; it requires a fundamental shift in consciousness. Conversely, each authentically new poetic line is the outward sign of a transformed mind behind it. Little wonder that critics, fellow poets, and later commentators have seized on these Williams quotes as encapsulating his legacy. Allen Ginsberg, a protégé of Williams, repeatedly invoked “a new world is only a new mind” as a Williamsian dictum when urging poets to find their own voice. Even outside poetry, educators have adopted Williams’s line to express how learning can alter one’s worldview: “A new world is a new mind,” as one literature professor summarizes, echoing Williams.
The endurance of this quote in such contexts speaks to its elegant truth: it names the nexus between creativity and thought. Williams’s words challenge us to consider that each line break — each act of form — represents a potential break with old habits of mind. And in that breakage lies the possibility of the new.
For Williams, the poetic line was not a mechanical unit of meter or a visual whim — it was an extension of cognition and perception itself. He conceptualized the line break as a mental breath, a pulse of consciousness on the page. In his practice of free verse, each line became a unit of attention, reflecting a single thought-beat or sensory impression before the mind shifted or “breathed” to the next. This is evident in the taut, enjambed lines of his shorter poems — for example, “The Red Wheelbarrow” unfolds in slender lines, each line isolating a simple image or word, making the reader’s mind dwell momentarily on “so much depends” / “upon” / “a red wheel / barrow” etc.
The line breaks slow down and shape the cognition of the reader, mirroring how the poet’s own mind momentarily fixes on each element of the scene. Williams once noted that his free verse lines were made “tense” by enjambment, which forces a pause or jolt at each break. These little jolts are cognitive: the break in the line enacts a break in thought, a tiny recalibration of what the poem is “about.” In this way, Williams treats the line as analogous to a thought or a perception: when a new line begins, the mind must start fresh, refocus, potentially change direction. “A new line is a new mind” is thus a literal description of his technique — the poem’s form perpetually renews the thinking process.
This idea became more explicit as Williams experimented with variable line structures later in his career. He developed the triadic, stepped line (seen in poems like “Of Asphodel” and “The Descent”), where longer lines are broken into three indented segments. Each segment is a phrase or image-unit, and the visual stepping creates a syncopated rhythm of thought on the page. Williams described this innovation as a “solution of the problem of modern verse,” giving flexibility while preserving a new kind of measure. Underlying that solution was his belief that form must follow the mind’s movement. Traditional meters, in his view, imposed an external order that no longer matched how modern minds actually worked or spoke. By contrast, his variable foot was meant to record the actual cadence of thought, which speeds up, slows down, stops and starts — much like his broken lines.
In Paterson, he writes:
“to make a start, / out of particulars… / ‘no ideas but in things’… / and the minds of men….”
suggesting that the assembly of particulars in new forms is how the mind creates meaning. The material of thought (ideas, perceptions) and the form of the poem (lines, rhythms) are “all apiece,” as he later put it. They are cut from the same cloth. If the mind leaps, the line leaps; if the mind lingers, the line elongates or repeats. Even the irregular collage structure of Paterson — with its sudden prose inserts and quotations — mimics a mind absorbing and recombining disparate materials in search of sense.
Williams was deeply influenced by developments in art and psychology that suggested process was paramount: he admired modern painters who fractured perspective (Cubists, Dadaists) and he knew of Freudian ideas comparing the work of the mind to a dream or collage. Accordingly, he envisioned the poem as a field of action for the mind, a dynamic space rather than a static form. Meaning, for Williams, emerges in the motion of the mind-as-poem.
Crucially, this meant that novelty of form had to be earned by novelty of thought. Williams rejected the notion of empty experimentation; a new pattern of line breaks was not worthwhile unless it corresponded to a genuinely new way of perceiving something. In Paterson II, when he asserts the need for a “new mind” to achieve a “new line,” it comes after passages describing the staleness of language and the bombardment of media — the noise of modern life that dulls perception. The cure, in the poem, is a mental revolution: to see one’s surroundings with fresh eyes so that each line of poetry cuts like a live wire.
We might say Williams viewed form as the trace of discovery. Each break in Paterson maps a breakthrough (or at times, a breakdown) in the poet’s understanding of his city and self. Small wonder that Paterson opens with the image of a waterfall’s spray and roar — a natural flow broken into mist — an analogue to language and thought being shattered into poetic lines. The fragmentation becomes constructive: out of the broken old, the poem recombines pieces into new wholes.
Williams’s contemporaries like T.S. Eliot and Ezra Pound often reached for mythic structure or historical allusion to organize their modernist epics; Williams instead relied on the living energy of the local mind at work, trusting that an authentic new structure would emerge from the act of attention itself. His “measure” is intimate and spontaneous, what one critic later called “dysfluent poetics,” even influenced by Williams’s experience with neurological patients (he was fascinated by how speech breaks down and rebuilds in aphasia, revealing the mind’s wiring). Thus, for Williams, the poetic line is a unit of both thought and perception — a concrete manifestation of how the mind navigates the world. To change the line is to change the thought pattern; to change one’s thinking is to necessitate a new line. This principle set him apart as a true modernist revolutionary of form: he achieved a fusion of content, consciousness, and craft that continues to influence poets to this day.
Williams’s intuition that “a new line is a new mind” finds an intriguing parallel in the way large language models (LLMs) generate text. At first glance, a mid-20th-century poet and a 21st-century AI might seem worlds apart, but both engage deeply with how context and continuity produce meaning. Where Williams spoke of the poet’s mind moving from line to line, an LLM like GPT-4 moves from token to token, each step a kind of “new line” in an ongoing composition.
Under the hood, the AI’s generation process is a traversal through a high-dimensional latent space — essentially an abstract vector space encoding concepts and relationships learned from human language. In this latent space, meaning literally has a geometric structure: words and ideas are mapped as vectors, and directions between vectors correspond to relations in meaning. For example, the direction from “small” to “large” represents the concept of growing in size; if the model takes a vector step in that direction from “mouse,” it moves toward “elephant” (a larger animal). In other words, the AI’s “mind” understands a new idea by moving in a new direction.
Each generated word is chosen by considering its context — the sequence of previous words — and finding a next word that lies in a semantically coherent direction from that context. This process is driven by the model’s attention mechanism, which dynamically weights the relevance of prior words to predict the next. Through contextualized attention, the LLM effectively “decides” at each step what aspects of the preceding text are most important, and this focus shifts as new words (new mini-lines of thought) emerge.
The result is strikingly akin to a poet following the momentum of thought: at each line break, the poet subconsciously asks, “What do I pay attention to now? What image or idea comes next?” The LLM does the same, albeit with matrices and probabilities — it re-contextualizes at each step, ensuring that what comes next is both connected to what came before and yet introduces something novel within the established context.
Critically, novelty in LLM output arises from moving into less-traveled regions of latent space while maintaining coherence. When an AI produces what we might call a creative or unexpected line, it is essentially plotting a course through latent space that connects familiar points in new and unprecedented ways. It might take two concepts that are usually far apart (say, “castle” and “octopus”) and, guided by the prompt and its internal knowledge, find an intermediate path that joins them in a meaningful sentence (perhaps describing an octopus as a castle’s guardian, for instance).
In doing so, the AI is engaging in algorithmic invention. One analysis describes this as traversing to new vectors in the latent space — a process identified as a form of algorithmic creativity, analogous to human creativity which often involves making unfamiliar combinations. Because an LLM has no fixed script or template (just as Williams refused to have a fixed form), it relies on emergent structure. The text’s form and direction unfold dynamically from the interplay of what has been written so far and the model’s vast learned associations.
Meaning becomes a kind of directional movement: the model knows that continuing in one semantic direction will yield a logical completion, while a twist in direction can yield surprise (yet still plausible) developments. This is reminiscent of Williams’s own compositional strategy in Paterson, where the poem’s “narrative” is not pre-plotted but emerges via associative leaps — guided by what the poem finds important in each moment (be it a piece of local history or a fragment of conversation).
The notion that “the mind and the poem are all apiece” also resonates here. In Williams’s case, it meant the poem is an extension of the poet’s mind. For an LLM, we might say: the generated text and the model’s “mind” (its state of activations) are all apiece. At each step, the latent vector state of the model (a kind of abstract representation of what has been said so far) directly informs the next word. The state is the synthesis of context, much as a poet’s state of mind (holding images, emotions, intentions in suspension) informs the next line they write.
Both systems — human poet and LLM — exhibit what we can call contextualized generation. They are fluid in form: a poet like Williams can slide from free verse to prose if the thought calls for it, just as an AI can shift register or style mid-text if prompted. And both demonstrate emergent structure: a poem’s shape or an AI story’s plot may not be fully apparent at the start; they become clear only through the act of creation, one line or one token at a time.
To put it metaphorically, Williams’s poet stands before the blank page much like an AI at the start-of-sequence token: armed with a vast internalized “training” (the sum of their language and life experience) and an initial impulse or prompt (an emotion, a scene), and then making the path by walking it. Each poetic line is a step in conceptual space; each step slightly alters the landscape of meaning to be navigated next. The AI similarly has learned an entire language distribution (its “training data”) and, given a prompt, starts generating, one word at a time, each word altering the context vector that the next must respond to.
If Williams insisted that “unless there is a new mind there cannot be a new line,” we might say of the LLM: unless the model shifts into a new region of latent space, it cannot produce a truly new continuation — it will otherwise loop on high-probability, cliché phrases (the AI equivalent of “recurring deadliness” in language). Indeed, one of the known failure modes of language models is getting stuck in repetitive loops; the remedy is often to adjust sampling parameters to encourage more exploration of less obvious options — in effect, to push the model’s mind toward the new line. Novelty emerges when the model dares to deviate from the most expected completion and instead ventures into the semi-unknown (while still guided by learned patterns for coherence). This is analogous to a poet breaking with expected syntax or imagery to create a startling line that nonetheless “works.” Both acts involve risk and intuition within constraints.
The comparison between Williams’s poetics and LLM text-generation ultimately illuminates shared principles of contextual creativity. In both, attention to context drives the next move, and by shifting this attention, something unexpected can arise. Williams the poet pays fierce attention to the world around him — the “things” — but also to the words he has already placed on the page. He lets the context he’s established suggest the path forward: a phrase in one line might trigger an association that becomes the germ of the next.
This is a kind of mental attention and feedback loop, not unlike the Transformer model’s self-attention mechanism where each token looks back at prior tokens to decide what comes next. The Paterson line “No ideas but in things” can be paralleled by an AI’s modus operandi: no next word but from preceding words. Both scenarios enforce that meaning is context-dependent. And in both, fluidity of form allows the work to adapt to the material. Williams’s writing could turn on a dime — lyrical in one moment, colloquial in the next — as he responded to the flow of thought and perception. Likewise, a modern LLM can genre-shift or change tone mid-text in response to a prompt twist, demonstrating form flexibility in real time.
Where this analogy becomes most thought-provoking is in the realm of emergent structure. We traditionally think of a writer as having some plan or intention that governs their piece. Williams complicates this: in a poem like Paterson, the structure feels organic, growing outward by accretion and surprise rather than top-down design. Similarly, when an AI produces a coherent essay or story, there is no central plan laid out by the machine — the coherence is an emergent property of many local decisions that statistically and semantically align.
There is an almost improvisational quality to both processes. We might recall that Williams was influenced by jazz and the improvisational rhythms of American speech; he sought a “swing” in his lines akin to music’s spontaneity. LLMs, while deterministic in their machinery, effectively improvise within the constraints of probability, weaving familiar motifs into novel combinations on the fly. One could say the LLM improvises meaning by traversing its latent space in an unguided yet not unguided way — probabilistic traversal within a learned possibility space. A recent analysis of AI art described it as “a probabilistic traversal through aesthetic possibility space, constrained by patterns learned… but not bound by human intentionality.” This perfectly echoes Williams’s artistic ethic: use the patterns of language and life, but do not be bound by old intentions — let the new emerge by exploring uncharted paths.
Finally, both Williams’s poetics and LLMs highlight the role of defamiliarization in creativity. The Russian formalist Viktor Shklovsky argued that art’s purpose is to “make the stone stony again” — to make us perceive freshly by defamiliarizing the familiar. Williams’s fresh lines, sprung from a “new mind,” do exactly that: a red wheelbarrow glazed with rain becomes a revelation; a simple apology note about stolen plums becomes a sly, poignant poem. By breaking syntax and line in unaccustomed ways, Williams forced readers to see slowly and experience language anew.
Intriguingly, AI-generated content often has a similar effect. Because an AI lacks human common sense and context, it can describe things from odd angles that a person might never think of. As one commentator noted, an AI approaches description from angles no human writer would naturally take, yielding a shock of recognition-in-strangeness. The AI knows the patterns but not the meaning — a radically non-human perspective — and so it might combine words in ways that defamiliarize even a cliché subject. In these moments, the AI-generated text achieves what Williams and the modernists strove for: it makes us see an old thing as if new. Both are leveraging the estrangement effect: Williams by consciously reinventing poetic form through new minds, and the AI by inadvertently offering an outsider’s recombination of our cultural artifacts.
Of course, there are profound differences. Williams’s “new mind” is tied to lived experience, emotion, mortality — the whole human drama — whereas an LLM’s “mind” is an engineered abstraction, with no intent or sentiment. Yet, as strange as it sounds, the output of each can converge in the realm of artistry. Williams wrote that “the mind, the poem… are all apiece,” suggesting a unity of being and expression. With AI, we have an expression without a being — all poem, no person. And yet, the structures it creates can mirror those of human creativity because they arise from the same linguistic building blocks and patterns of association. It is a testament to the power of context and structure that even a soulless algorithm can produce lines that sometimes move or startle us. We might say the context itself has a kind of logic and momentum that, if properly engaged, yields meaningful form.
Both Williams’s poetics and LLM writing thus underscore an essential truth: attention to context coupled with freedom from precedent can yield the genuinely novel. Whether it is a poet paying attention to the shape of thought in a new way, or a neural network attending to patterns of words, the emergence of something never-before-seen (a new line, a new insight) requires stepping beyond the already known. Williams, as a modernist, championed attention, precision, and courage in breaking from tradition. Today’s AI, in its alien way, exhibits a parallel courage — it has no concept of “making sense” or “sounding conventional” beyond statistical likelihood, so it can easily tip into strange phrasings that, when they succeed, feel like poetic leaps. In both cases, surprise is a sign that the frontier of the mind or model is being tested. Williams often surprised his first readers — what to make of these oddly broken lines about chickens and wheelbarrows? — but those willing to adapt their reading mind found new beauty and clarity there. Likewise, AI outputs can surprise us into laughter or reflection, showing us combinations of our own language we didn’t anticipate.
“A new line is a new mind.” Williams’s adage carries a performative force — it does not just describe, it provokes. It invites writers and thinkers to approach each line as an opportunity to reinvent one’s understanding. In the lecture-hall of modern literature, it resounds as a call for perpetual creativity: do not settle into old patterns, for each line can begin the world anew.
The journey we’ve taken, from Williams’s Paterson falls to the transformers’ latent space, reveals that this principle transcends its original context. In poetry, a new line can be a leap of insight or image that reconfigures what the poem is about. In the realm of AI, each new “line” of generated text is a vector move that can open up novel semantic territory. Both processes hinge on contextualized attention — the ability to carry forward what matters and leave behind what doesn’t, to break when needed and flow when needed. Both rely on a fluid interplay of continuity and disruption. And both attest to an almost paradoxical truth: through constraints, freedom; through form, discovery. Williams found freedom by casting off inherited forms and listening to the natural cadences of language and thought. LLMs, constrained by their training data and algorithms, nonetheless generate unplanned creations that can surprise even their makers — a different kind of freedom arising from the absence of human preconceived intent.
In a performative sense, one might imagine William Carlos Williams and a state-of-the-art language model in a kind of call-and-response across time. The poet asserts, “The mind, the poem, are one”; the model responds, in its billions of weighted connections, by demonstrating how form and “mind” (if only a statistical mind) indeed converge in output. The poet says, “No ideas but in things,” and the model, trained on innumerable things described in text, dutifully fills its verses with concrete nouns and scenarios drawn from the human world. The poet declares, “Unless there is a new mind there cannot be a new line,” and the model — devoid of any true mind — curiously validates the inverse: that by producing a new line, it can sometimes simulate a new mind behind it. The human reader stands between them, witnessing how context begets form in each case, and how novelty sparkles at the intersections: between perception and utterance, between prompt and completion, between what has been and what might be.
In the end, what Williams championed was attentive human consciousness — the ever-renewing mind as the engine of art. That remains as vital as ever. Our technologies may evolve to generate fluent poetry or simulate dialogue, but the truly new — the line that changes literature or changes one’s life — still likely springs from a living consciousness daring to see differently. Williams, the good doctor of Rutherford, New Jersey, healed American poetry by prescribing it fresh eyes and unshackled lines. “A new line is a new mind” was both his prescription and his prophecy. It suggests that each line of a poem (or indeed, each sentence we speak) carries the potential of a tiny revolution — a chance to reimagine the world.
The lesson we take, whether crafting verses or coding algorithms, is to approach creation not as the reiteration of safe formulas, but as an adventure into the unknown edges of sense. As Williams put it in Paterson: “— Say it, no ideas but in things —”. The things change when the mind engaging them changes. So may we embrace the new mind at the break of each new line — and in that ever-renewing breakage, glimpse whole new worlds.
William Carlos Williams’s Paterson and Pictures from Brueghel (“To Daphne and Virginia”); Williams’s essay “The Poem as a Field of Action”; commentary by Eliot Weinberger; modern analysis of latent space in AI; and the concept of defamiliarization in AI art discourse.
The interwoven citations illustrate the dialogue between Williams’s words and contemporary understandings of creativity, human and artificial. Each citation, like each line, is an entry point into a larger context — a reminder that every line of thought, to be truly new, must conversely be in conversation with what came before, yet daring enough to break away.