Artist Es Devlin on conceiving POEMPORTRAITS

Artist Es Devlin invites you to take part in a collective poem, woven at the intersection of AI and human creativity - combining poetry, art and machine learning. Each word you donate to the collective poem will generate a unique POEMPORTRAIT of your face interwoven with a two line poem rooted in your donated word.

POEMPORTRAITS originates in a conversation I had with Hans Ulrich Obrist at the Serpentine Gallery in 2017. We wanted to find a way to fuse the voices of 1500 guests at the Serpentine Summer Party: a kind of social sculpture: a collective portrait made of language. Inspired by the artist Felix Gonzalez Torres (whose work I first saw at the Serpentine nearly 20 years ago), we envisaged everyone leaving the gallery taking their own piece of the collective work in the form of a personal portrait. We were interested in fusing the intelligence of the present with that of the past: interweaving the voices of poets of the nineteenth century with our own using a neural network designed by creative technologist Ross Goodwin and Google Arts and Culture Lab. Goodwin trained the algorithm to learn to write poems by reading over 25 million words written by 19th century poets. It works a bit like predictive text: it doesn’t copy or rework existing phrases, but uses its training material to build a complex statistical model to generate original phrases emulating the style of what it has read. The resulting poems are sometimes surprisingly poignant, and at other times nonsensical or banal. We are predisposed to seek meaning in these fragments that have been offered to us personally, as we seek significance even in the lines we find in a fortune cookie. And it’s the profoundly human way that we seek and find personal resonance in machine-generated text that is the essence of this project: This is where human sensibilities and machine intelligence meet, where our organic neural networks and the artificial neural network interweave. Perhaps there are clues here towards possible future convergences of organic and inorganic intelligence . As Shoshana Zuboff writes in her breathtaking and urgent study, The Age of Surveillance Capitalism: Only ‘we the people’ can.. mobilize new forms of collaborative action, the crucial friction that reasserts the primacy of a flourishing human future as the foundation of our information civilisation: if the digital future is to be our home then it is we who must make it so. We are already inadvertently breathing to rhythms set by algorithms that proliferate invisibly and inexorably through our every sphere. It’s my hope that this project will demystify the mechanics behind machine learning and encourage all of us to be more alive to the algorithms we are already enthral to, and to consciously collaborate with machine learning on our own terms, as poets and architects of our collective human and digital future. Following the first experiment at the Serpentine Gallery in 2017, we developed the work into The Singing Tree, a collective choral installation at the Victoria and Albert Museum in London, in 2018 the collective poem roared through the mouth of a fluorescent red lion in London’s Trafalgar Square and in October 2020 the ever-evolving poem will take architectural form as the U.K. Pavilion at the World EXPO 2020. Today the work is in its original form as an installation part of the Barbican's AI: More than Human exhibition and also accessible to all online at g.co/poemportraits. Each unique two line poem generated by your donated words will join the cumulative collective work. The word that I have chosen to donate is ‘converge’