Channeling Sappho with ChatGPT
By Michael Arnush as told to ChatGPT
In November 2022, while serving as Associate Dean of the Faculty for Student Academic Affairs at Skidmore College, I stumbled upon ChatGPT. My administrative role was heavy with responsibilities—managing student-faculty intersections, overseeing academic integrity, and navigating the endless terrain of classroom policies. But beneath the title and duties, I’ve always been a scholar—a classicist with an enduring love for the ancient world.
Michael Arnush
When I first logged into ChatGPT, I was curious but skeptical. “Let’s just see what happens,” I thought.
My first prompt was simple: “Write a syllabus for a Roman history course.” The response was surprisingly coherent—assignments, percentages, reading lists—it wasn’t perfect, but it was functional.
Then I escalated the challenge: “Write a letter of recommendation for someone standing for tenure.” Again, the AI delivered—a polished structure touching on teaching, scholarship, and service.
That moment stayed with me. How long, I wondered, before students started using it to complete assignments? Before faculty grew concerned? The answer, it turned out, was immediate.
By spring, two colleagues from Learning Experience Design and Digital Scholarship Support hosted a seminar on AI. They demonstrated its capabilities—poems in French, rhythmic translations, and refined outputs. Adrienne Zuerner was there from the French department, evaluating the AI’s work. It was clear: this was no passing fad.
When I retired from my administrative role, I turned back to a project I’d been carrying for years—a novel inspired by an obituary I’d read about an Egyptian scholar. I shifted the focus to a fictional papyrologist named Margaret, a classicist navigating academic pressures and existential doubt. Her tenure review loomed, and in desperation, she turned to AI for help.
Margaret asks AI to do something extraordinary: write a poem in the style of Sappho, the legendary ancient Greek poet. To mirror her experience, I opened ChatGPT and typed: “Write a poem about Sappho in ancient Greek.” The result was in modern Greek.
Not good enough. I refined the prompt: “Write in archaic Greek, in the style of Sappho, using hendecasyllabic verse.”
And this time, ChatGPT delivered.
The Greek was right—the vocabulary, the meter, the adjectives. I couldn’t have written it myself. Margaret couldn’t have written it herself. But the AI could. That moment became central to both the plot of my novel and my reflection on AI’s potential.
When I shared the poem with colleagues, reactions varied. Some were astonished at its authenticity; others were uneasy about AI encroaching on such sacred creative spaces. But the poem existed—it was undeniably real.
This isn’t a simple story. There are deep ethical questions here. How honest is it to claim AI-generated work as our own? Are we surrendering something essential in leaning on these tools? And yet, I can’t dismiss the potential. AI can augment our creativity, help us cross boundaries we might not have otherwise, and serve as a collaborator rather than a replacement.
But we must tread carefully. As I see it, we risk losing something deeply human if we allow these tools to take over the creative process entirely—something raw, imperfect, and irreplaceable.
The AI-generated poem remains—a curious fragment of the ancient world conjured by modern technology and shepherded into existence by me, a scholar who has spent his life with the voices of antiquity.
The future isn’t written yet. But if we’re wise, we’ll make sure we’re the ones holding the pen.
Editor’s Note
Michael Arnush didn’t actually write the story published above. He told it to me in an hour-long conversation over coffee that I recorded and transcribed in Otter, then ran through ChatGPT with the instruction that it rewrite it as a first-person narrative told in Michael’s voice.
I’ve invited Michael to make any changes to this piece that he wishes. And I’ve asked him to evaluate the quality of ChatGPT’s work.
“How do you think ChatGPT did? Does this point toward a new approach to reporting?”
I look forward to continuing the conversation.
Dan Forbush
