There is a new thing you can do on the Internet with machine learning. Poem Portraits is an experimental art project by artist and stage designer Es Devlin in collaboration with Google Arts & Culture. You only enter a word into their generator, snap a photo, and you get a piece of poetry back for you to keep or share.
What do I have to do?
You just visit their project website on a desktop or mobile browser. Once everything is loaded up you think of a cool word and enter it into the text field. You then “donate” this word into the art machinery and it begins to process the data. Or you keep watching the wall of previously created poems scroll through the screen for as long as you like.
The next step is a quick screen about how the system has achieved poetry skills by learning from over 20 million words of 19th-century poetry. You just click “continue” to go on. Now you give the system your agreement to access your webcam in order to make a photo of you. It can be that your browser will also double-check this with you then so you basically click or tap on “authorize access” and agree on any prompt that might pop up.
If you don’t want your face to go into the system, you can also skip the photo, as it is not required in order to generate your poem. I liked to try it with a camera, so excuse the silly selfies used here to display the results of my test. So – now everything is ready and loaded up. You get your photo, if you chose to make a selfie, along with some cool filters and an overlay of text, being the poem that was created for you. Underneath you get the poem in clear writing with the option to save it, read collective poems, or retry.
Save it, share it, print it, have fun!
Fun, isn’t it? Now make sure to save your cool shots and poems with your family and friends, on social media or wherever you like. I could also imagine that it’s fun to print your best ones out and stick them to the wall. You could even photograph your pets or friends instead. What do you think? Try it out here.
YouTube: Poem Portraits – Poems generated with machine learning by Es Devlin
Photo credit: Examples were created by Christopher Isak of TechAcute on the Poem Portraits art project website.