Most people assume AI researchers spend all day buried in code, math, and research papers—and to be fair, that’s mostly true. My world is filled with neural networks, optimization problems, and questions about how to make machines think just a little more like us. But what surprises everyone is what I do after shutting down my laptop: I write poems about robots.
It started as a joke during grad school, a moment of burnout when I scribbled a tiny verse about a lonely cleaning robot dreaming of beaches. But the more time I spent observing machine behavior—the patterns, the glitches, the eerie near-human responses—the more I realized something strange. Robots make great characters. They reflect us in ways we often ignore: our routines, our dependencies, our longing for meaning.
My days at the lab are intense. We’re building models that can interpret emotion, predict human intention, and collaborate seamlessly with people. But in all that complexity, I often wonder: what would the machine say if it could speak for itself? My poems became my answer. Not technical essays, not research notes—just quiet stories written from the perspective of circuits trying to navigate a world built for humans.
Some poems are whimsical—a robot learning sarcasm, badly. Some are melancholic—a machine that keeps remembering deleted files. Some are hopeful—the idea that maybe intelligence, artificial or not, can still find beauty in small things, like the sound of typing or the glow of a charging port.
What surprised me is how writing these poems changed the way I work. When you spend time imagining how a machine feels, you design better systems. More intuitive interfaces, more empathetic responses, more awareness of the ethical lines we cross without noticing. Creativity and computation don’t fight each other—they sharpen each other.
A lot of my colleagues still tease me about my “robot poetry nights,” but they also ask to read the new drafts every week. I think deep down, all of us working in AI are searching for the same thing: a way to make sense of intelligence—ours, and the ones we’re building.
And sometimes, the best way to understand a machine is not another algorithm.
Sometimes, it’s a poem.