It’s body consists of 3D printed and precise laser-cut materials. Treads allow it to be fully mobile and self-propelled. An array of sensors enable it to detect temperature, wind speed, sound, and other environmental conditions, and thanks to machine learning, it can operate free of human intervention. This isn’t some autonomous sentry vehicle for the military or a robot designed to work in dangerous conditions, however: It’s actually a poet—or, at the very least, it’s a machine that writes poetry.
Designed by Yuxi Liu, a graduate student at the University of Edinburgh, the affectionately named “Poet on the Shore” is a small robot that can meander along beaches and scrawl poems in the sand based on its surroundings, as if R2D2 decided to move to the seaside and take up verse.
“The robot has a number of sensors that enable it to sense the world around it: the sea, the wind, the sounds etc.,” Liu explained in a post on her website. “Empowered by machine learning, it can discover the patterns, and create associations in its mind.”
Although Liu’s “Poet on the Shore” is perhaps the most novel example, it fits into a broader trend in which scientists and technocrats have begun to develop artificial intelligence (AI) programs that can write poetry. These developments are an exciting sign of what technology can do, although they have drawn criticism from poets who feel that AI lacks the nuance and complexity that give meaning to true poetry.
Jack Hopkins, a British computer scientist, created an algorithm-powered AI that can pen poems according to specific rhymes, rhythms, or themes. After some light reading—seven million words of 20th-century English poetry drawn primarily from poetry books available online—the algorithm was able to write poems of its own by stringing together words within lines or, for rhymes, across lines. It would check itself after initially writing to ensure that content within sections of the poem was similar enough to go together; if sections did not pass the content check, then the AI would rewrite the section.
Some readers were unable to distinguish between poems authored by Hopkins’ AI from poems written by real people; you can guess for yourself using this test Hopkins developed to study the AI’s. On the other hand, the poet Rishi Dastidar criticized the program’s poetic abilities, arguing that it could only imitate the works and styles of humans and that the algorithm produced poems that were all style and no substance.
Dastidar raises a compelling point: How can an emotionless machine write real poetry? After all, poetry has long been recognized as a fundamentally human pursuit, so the sudden appearance of robo-bards calls our very ideas about humanity into question. Meanwhile, Liu—invoking the nascent robot rights movement—suggests that a knack for poetry could actually expand the category of humanity to include robots and AI. So while the question of how well they write or how human machine poets truly are may go unanswered, the debate over the issue reveals an important truth about poetry: No matter the author, poetry can help us to clarify and think more deeply about what it means to be human.