PODCAST/ ESSAY: What Happens When AI Writes Better Poetry than Us? (aka Poets Gotta Poe!)
Machine Cinema speaks to AI Policy Expert/ Designer Jamie Joyce
Links to longer conversation on Youtube, Spotify, and Apple. Follow us on Instagram.
"The machines are coming for the poets now, too."
You might hear this from any number of creative professionals grappling with AI's rapid evolution, but it takes on refreshing depth coming from Jamie Joyce. As founder of the Society Library, Joyce spends her days cataloging policy arguments, mapping government documents, and building comprehensive databases of competing perspectives on complex societal issues – work that gives her a uniquely informed view on both AI's capabilities and its limitations.
Yet when a widely-shared Nature study claimed that "AI-generated poetry is indistinguishable from human-written poetry and is rated more favorably," Joyce's response revealed another dimension to her expertise. Beyond her work in policy analysis, Joyce is also a former slam poetry organizer who once ran the Sunburnt Poet Society, allowing her to bridge technical understanding with genuine artistic experience.
The study that sparked her interest found that when 1,634 participants were presented with both human and AI-generated poems, they identified AI poems as human-authored more often than real human poetry, with an accuracy rate of just 46% – below random chance. Even more striking, AI-generated poems received higher ratings for qualities like rhythm, beauty, and emotional impact. This empirical finding stands in direct tension with Joyce's passionate response:
"For me, what a poem quintessentially is, is just a form of human expression," she reflects. "And so in that way, AI-generated poetry can never be as good unless it is human."
The contradiction is illuminating. While the average reader may prefer AI-generated poems when judging purely on aesthetic qualities, Joyce argues that poetry's value isn't reducible to these surface characteristics. Her perspective challenges us to consider whether we've been asking the wrong questions about AI creativity. Perhaps the comparison isn't about which produces "better" poems according to traditional metrics, but about fundamentally different kinds of value – one based on technical execution and another based on human connection.
The Value of Constraint
The most compelling insight Joyce offers is deceptively simple yet profound: "To be human is to be limited."
This idea emerges naturally from her work at the Society Library, where she's pioneered systems that map diverse perspectives across complex policy issues. "We talk a lot about human bias," she explains. "To be human is bias because we get asked all the time, 'How do you overcome bias in your work?' And it's like, well, there's no such thing as unbiased. To be human is bias."
Rather than viewing our limitations as weaknesses, Joyce suggests they might be the very source of what makes human expression meaningful. When we contemplate older creative works—whether music, literature, or visual art—part of their power comes from understanding the constraints under which they were created.
"When we think about music through the 20th century, they only had access to so much technology. The recording can only be so good. With what they had, this is them, just as constrained as humans are in their time and talent and attention and resources. This is that expression that came forth."
This reframing of limitation as a feature rather than a bug resonates deeply in creative technology circles. Constraints often drive innovation – whether it's the limited color palette that defined early video games or the time constraints that power entire genres like the 48-hour film festival. Joyce's formulation gives philosophical weight to what many creators intuitively understand: sometimes the box is what makes the thinking outside of it so interesting.
The paradox emerges: AI can produce work that meets or exceeds human output by certain metrics precisely because it isn't bound by human limitations. Yet something essential might be lost in that unlimited capability.
The Human Connection
Joyce illustrates this through her experience reading Charles Bukowski's poetry: "I like Bukowski because I like thinking of him as a person. When he is writing a poem where he's looking out a window, I enjoy picturing him as this gruff, old dog, still writing lusty poetry."
This connection—knowing there's a human life and consciousness behind the words—transforms how we experience art. "It's not just the experience of how you consume it, but how you experience it and knowing there's another human on the other end of that," Joyce emphasizes. "It's like having a conversation with a person."
The distinction becomes clearer when considering our reaction to learning something was AI-generated rather than human-created. "You could be presented a song that's actually AI-generated and sold to you as this undiscovered, previously unreleased song, and you may be so excited because you love the artist," Joyce explains. "And then as soon as you're told it's actually AI-generated, there is that something suddenly missing, right? There's like just a little vacuum, a little gap, a little heartbreak that happens."
The Slop Paradox
Yet here lies the central paradox of the Nature study: people were perfectly happy with AI-generated poetry—even preferred it—until they learned its origin. What some AI commentators call "slop" points to this reality—there exists a vast universe of content that people consume and enjoy without concern for its origins.
Music provides the most established example of this phenomenon. Electronic dance music has evolved far from human performance with traditional instruments to become almost entirely digital—a space where machine generation and human curation blur. In clubs around the world, dancers respond to the atmosphere and emotional resonance rather than contemplating the human craft behind each track. The experience prioritizes the immediate sensation—the "vibe"—over any connection to a specific human creator. This reality has existed in music for decades before the current AI revolution touched other creative fields.
This tension between Joyce's intuitive defense of human expression and the empirical data showing people can't distinguish (and often prefer) AI-generated content raises profound questions. If human limitations are so valuable, why don't they consistently produce results that audiences recognize and prefer? If knowing something is AI-generated changes our perception, what does that reveal about how we actually value creative work?
Evolving Frameworks
Perhaps what the Nature study and Joyce's reaction reveal is not a contradiction but a clarification of what we actually value in creative works. As AI systems become increasingly sophisticated, our frameworks for understanding creativity must evolve beyond simple metrics of aesthetic quality.
"I think that framework is probably coming," Joyce predicts. "It's probably coming from the field and the practitioners themselves."
In film—a medium that "quintessentially requires technology"—these distinctions become especially complex. "I think that boundary is probably gonna be a little bit more blurred," she notes, imagining a future where creative works might come with metrics: "60% AI, there was like 40% of this human intervention."
What's emerging isn't a simple hierarchy of quality but a more complex matrix where different modes of creation serve different purposes. Some content is consumed for pure aesthetic pleasure or functional enjoyment—areas where AI excels. Other creative works are valued specifically as human expressions, as artifacts of connection across consciousness—a domain where human limitations become features rather than bugs.
Joyce remains optimistic about AI's role in enhancing rather than replacing creative experiences. "AI actually speeding that up, I think could really contribute to a much more rich enjoyment of life, including all of the arts," she suggests, while acknowledging the enduring appeal of the creative journey itself: "There will always be people who love the pursuit."
The resolution may lie in recognizing that we're witnessing a fundamental reconfiguration of value in creative work. Just as photography once freed painting from pure representation, allowing it to explore new dimensions of expression, AI may be freeing human creativity from certain technical constraints while simultaneously highlighting the unique value of human-to-human connection.
As the creative landscape evolves, Joyce offers a closing thought that serves as both caution and celebration: "As we march into this AI age, let's look to our brothers and sisters and realize that there is a fundamental part of them that needs to be expressed that should also be protected. Painters gotta paint, poets gotta poeticalize everything."
The most profound insight might be that AI isn't redefining what it means to be human—it's revealing what was always true. Our limitations aren't simply impediments to overcome, but the very source of connection we seek through art. When we know something was created by another limited, biased human consciousness reaching out across the void to connect with us, we experience something fundamentally different than when we engage with even the most aesthetically pleasing algorithmic output.
In this light, the Nature study doesn't undermine Joyce's thesis but amplifies it. The fact that our perception changes when we learn something is AI-generated demonstrates that what we're seeking isn't just beauty or technical mastery, but connection with another consciousness—the essence of what it means to be human.
To learn more about Jamie and her work follow her on Linkedin, X, and at her website.
Words by humans + Claude. Images by Sora and Canva.