When Good Becomes Too Good: The Curious Case of AI Poetry

When AI poetry outshines human efforts, does art lose its soul—or find a new one?

Reviewed by Margaret Foley

Poetry has always been a uniquely human endeavor, a medium where emotion and intellect merge to create something timeless. Yet, research is challenging this assumption. A new study reveals that AI is not only capable of crafting poetry indistinguishable from human-authored works but can sometimes surpass it in perceived quality. Participants in the study were asked to evaluate poems written by AI alongside those by acclaimed poets such as Sylvia Plath and Emily Dickinson. Their ability to discern the difference was worse than chance, and, intriguingly, AI-generated poems often scored higher in rhythm, beauty, and emotional clarity.

This revelation raises profound questions about creativity, artistry, and even the limits of human comprehension. If AI can write poetry that appeals so perfectly to us, is it truly creative—or merely a mirror reflecting our own biases and preferences? And what happens when art, in its pursuit of perfection, becomes “too good” for its own good? Take a deep breath and let's take a closer look.

The Allure of Simplicity

The study highlights what the researchers call the “more human than human” phenomenon. AI-generated poems, while technically less complex, were often mistaken for human creations because of their accessibility and emotional resonance. These poems were clear, rhythmic, and easy to engage with, capturing the immediate attention of readers. Human-authored poems, on the other hand, with their layered metaphors and nuanced themes, were more frequently misidentified as machine-generated.

This preference for simplicity may not be surprising in today’s world. In an age of rapid information consumption, where attention spans are fleeting, poetry that offers instant emotional gratification feels more relevant than the cerebral, ambiguous works of T.S. Eliot or Sylvia Plath. But there’s an irony here: The very simplicity that makes AI poetry appealing may also limit its longevity. Great poetry endures because it resists easy answers, offering layers of meaning that unfold over time. If AI poetry is “too good” at catering to modern tastes, does it risk becoming disposable?

The Role of Complexity

Traditionally, complexity has been a hallmark of great poetry. Works like Eliot’s The Waste Land or Dickinson’s enigmatic musings invite readers into a labyrinth of interpretations. These poems challenge us to grapple with ambiguity, rewarding careful attention with ever-deepening insights. But complexity isn’t always welcomed. For many readers, it can feel alienating, a barrier to emotional connection.

AI poetry, by contrast, leans toward clarity. Its simplicity is both its strength and its limitation. On one hand, it allows more people to connect with the work, democratizing poetry in a way that echoes the rise of accessible art forms like photography. On the other, it raises a critical question: Can poetry that lacks ambiguity truly reflect the complexity of human experience?

Here, the tension between simplicity and complexity mirrors a broader debate about art and technology. In science, complexity is often celebrated, yielding breakthroughs that expand our understanding of the universe. AI’s ability to process massive datasets has revolutionized fields like genomics and climate modeling, where incomprehensible patterns are decoded into actionable insights. But art operates differently. Its value lies not only in resolution but in its ability to provoke, to evoke, and to remain unresolved.

The Problem With “Too Good”

Perfection has its downsides, particularly in art. AI-generated poetry, while polished and precise, risks feeling sterile, its flawlessness devoid of the imperfections that give human art its soul. Part of what draws us to human creativity is its vulnerability—its capacity to surprise, fail, or veer into the unexpected. An AI poem, no matter how technically impressive, may lack the emotional resonance that comes from knowing it was born of lived experience.

The phenomenon of “too good” is not unique to poetry. AI-generated music, images, and even novels have faced similar critiques. Their outputs, while aesthetically pleasing, often lack the depth and unpredictability of human creations. This sterility becomes even more apparent when AI-generated art begins to operate in realms of complexity that exceed human comprehension. I've described this as AI’s “superreality,” where outputs are so advanced they elude our understanding entirely. What value does poetry hold if its meaning is inaccessible to its audience? Can it still be considered art, or does it become a curiosity of advanced computation that has meaning that is undecipherable by the human mind.

A Catalyst for Reflection

Yet, incomprehensibility is not always a flaw. Throughout history, groundbreaking art has often been misunderstood in its time. Think of Stravinsky’s The Rite of Spring, which caused riots at its premiere, or Picasso’s cubism, which shattered traditional notions of perspective. Both were initially dismissed as incomprehensible, yet they are now celebrated as masterpieces. AI poetry might follow a similar trajectory, pushing the boundaries of creativity and inspiring new ways of thinking about art.

Perhaps the greatest value of AI poetry lies not in what it creates but in what it reveals about us and how it drives human cognition and emotion. The study’s findings suggest that we favor accessibility, clarity, and emotional resonance—qualities that align with our need for connection in a fragmented world. But they also challenge us to reconsider the role of ambiguity and complexity in art. Are these qualities essential to creativity, or are they artifacts of human limitations that AI is helping us move beyond?

The Future of AI and Poetry

As AI continues to impinge upon our humanity, it raises critical questions about the nature of art and the role of technology in human creativity. Can poetry still be considered art if it is perfectly optimized for human preferences? Or does its value lie precisely in its imperfections, in its capacity to resist easy interpretation?

AI poetry, for all its brilliance, may ultimately be a mirror—a reflection of what we value, what we seek, and what we fear losing. But mirrors can also distort, offering not truth but a vision shaped by bias and expectation. As we look more closely at the possibilities of AI in art, we need to ask whether we are willing to be challenged by it, to grapple with its complexities and even its incomprehensibilities.

In the end, poetry—whether human or machine-made—remains an exploration of what it means to be human. And perhaps the greatest gift AI can offer us is the opportunity to reimagine that question, to see ourselves anew in the words of a machine.

O, what a tangled web we weave when first we practise to deceive!