Machine Learning Porn visualises an AI’s pornographic ruminations narrated by an AI generated non-binary voice.

Yahoo trained a neural network* to flag and censor sexual imagery from the internet. It learnt by being fed a dataset of thousands of graphic images, independently working out what they had in common to make them explicit. This neural network was then reverse engineered to create the most pornographic imagery from a computer's perspective.

Morphing between the yonic and the phallic, the images created are not gendered in a binary sense. They flow through the queer spaces between identity, gender, bodies and sexuality.

Previous work Machine Learning Porn (2016) exists as 5 physical editions and has been internationally exhibited.

*Not Suitable for Work (NSFW) classification using deep neural network Caffe models (Yahoo, 2016) and visualisations using Synthesizing the preferred inputs for neurons in neural networks via deep generator networks (Nguyen, 2016). Inspiration from Gabriel Goh, the machine learning researcher who originally discovered the pairing technique.

Jake Elwes looks for poetry and narrative in the success and failures of digital and AI systems, while also investigating and questioning the code and ethics behind them. His work has also been exhibited in museums and galleries internationally, including the ZKM, Gazelli Art House, Today Art Museum and Victoria and Albert Museum.