We understand the world through metaphor. Our minds seek and spin patterns and connections, likenesses and equations. Biologist and anthropologist Gregory Bateson observed that metaphor is “how the whole fabric of mental interconnections holds together. Metaphor is right at the bottom of being alive.” As above, so below.
The most effective and explicit specimens of metaphor are found in poetry. Weaving metaphors into poems is an age-old and far-flung human act: we see and search the world with a poetic mind.
Core concepts & related writing by project creator Corey Pressman:
Why, then, do we search a simple on-line image bank with such literal terms? Because the robots haven’t been taught our poetry. They only know the technical EXIF metadata and whatever descriptive adjectives they’ve been begrudgingly fed by underpaid (or unpaid) interns. But what if we write poetry for the robots? What if we used poetry and metaphor as metadata? Would a search for “eyes” return images of stars?
Poetry for Robots is a digital humanities experiment instigated by this Imaginary Papers blog post and sponsored by Neologic Labs, Webvisions, and Arizona State University’s Center for Science and the Imagination. Starting today, we will populate a database with poetic metadata affiliated with specific images.
Beyond this, we may extrapolate and investigate further. Will this reveal a “pattern of metaphors,” as posited by the great author and poet Jorge Luis Borges? Can an algorithm, informed by our poetic input, generate compelling works of its own? Let’s compose poetry for the robots and see.