Mirror Selfless dives into the stream of post-photographic practices in order to reflect on one of the uses that contemporary imagery has developed: to show ourselves. A data set composed of a thousand screen captures taken from Instagram’s #mirrorselfie tag is used to train a neural network model which task is to generate similar images. A limited time of training prevents the model from generating hyper-realistic results, therefore creating abstract compositions in which human remains and cell phones are intertwined and hardly identified. This work has won a federal prize at the BIT International Biennale of Art, Science and Technology, and has been selected both by the UADE Prize of Digital Arts and the ILDA Prize of Data + Art.