Individual Work
Poems About Things

Poems About Things is a web app developed by Danish artist Andreas Refsgaard. The work utilizes a machine vision model to perform object recognition from a phone’s camera, and then generates little poems about said objects based on the Google Suggest API’s attempts to finish generic search queries that contain the object’s name. That is to say, the poems are based on queries containing the word that the machine vision model returns for the items it is pointed towards; though sometimes accurate, the software often gives surprising and seemingly arbitrary results. As a personal example, I have frequently been identified as a ‘restaurant’. The poems – consisting of a number of stanzas that all take the form of generic search results – are printed on top of the image from the phone’s camera, and thus are taken out of the context of the Google search engine. Still, the stanzas retain a stylistic resemblance to Google search queries that is recognizable even without the search engine interface. As readers, we recognize this style of querying, and immediately relate it to a search engine-based quest to learn something about the world. The surprisingly earnest, and sometimes absurd, questions sustains a state of computerized wonder in relation to our physical surroundings. The queries are left unanswered; even though the Google search engine could doubtlessly retrieve a vast number of hits for each query in just a few splits of a second, it seems that these questions in particular are directed at us, the readers. Even though the queries are formulated as Google-ready inputs, Poems About Things is not looking for answers that are already there, on the indexed Internet, but instead the further questioning that arises in the specific situation of pointing the camera, processing the input, presenting the output, and reading the poems. As readers, we too begin questioning, rather than answering. Not only do we question what the machine sees, but we also get to join the machine in its curious and sometimes completely arbitrary questioning about that which it sees. The work, then, is not so much about objects – or ‘things’ – surrounding the phone as it is about machine vision, Google, and the very act of querying: it both troubles and poeticizes the proprietary software platforms which increasingly order the world. Poems About Google – or Google Queries About Machine Vision.

This entry has been reworked from: Erslev, Malthe Stavning. "Exhibiting, Disseminating, Teaching: Digital Literature in Danish Public Libraries." Proceedings of the 2020 ELO Conference and Media Arts Festival, 2020, https://stars.library.ucf.edu/elo2020/asynchronous/proceedingspapers/7/

Screenshots: 
Author statement: 
Poems About Things is a project that generates poetry from everyday objects around us. It consists of a mobile website that constructs quirky sentences about the objects it sees through the users camera feed. As the user's camera focuses on an object, a built in machine learning model gives its best guess as to what object it is seeing. Based on this estimate a short query is sent to Google Suggest API, which in return sends back a list of sentences inspired by the detected object. The poem appears on the screen overlaying the image of the object. It consists of a handful of sentences expressing thoughts, questions or comments related to the immediate object as well as the bigger world outside it.