New Media Arts

New Media Arts - Narratives

Manolis Perrakis & Mathis Antony / HK

Narratives is a set of reclining chairs that visitors can lay on while listening to machine-learning-generated advice on self-improvement. The model has been trained using a myriad of self-help and motivational articles and posts, along with content gathered from around the web. The reclining chairs have vivid upholstery that celebrates life but also portrays small details such as flies or pests eating rotten fruit to highlight the inevitability of death. The advice given to participants is influenced by an image captured of the participant’s face while they are reclining on the chair. The computer vision recognises the age, gender and even some emotional characteristics of the person’s face in real-time and uses them as seeds for the text generator. The advice will be a mix of spiritual, motivational and self-help influenced by the person’s face. Narratives is an exploration of machine-learning generative models and the future of the human-AI co-existence.

About Manolis Perrakis
Manolis Perrakis studied Fine Art at the Kunstakademie Düsseldorf, Germany. He has had various exhibitions, always using mixed media by bridging the gap between technology and traditional art methods. His artworks concentrate on playful aspects of social and political topics pushing the limits of the latest tech. He has been living in Hong Kong since 2009, where he co-founded the hackerspace Dimsum Labs and has worked for various advertising agencies as a creative technologist. Manolis describes himself as a part-time artist, wannabe engineer and dedicated hacker, who currently spends most of his time working as the Head of Creative Tech at Ogilvy Hong Kong focusing on building innovative products and services.

About Mathis Antony
Mathis is a HK based software developer. He quit his PhD in physics and has worked in machine learning since. He likes to build systems with a net positive effect on society. Sometimes he dabbles with arguably more artistic uses of machine learning.

Special thanks to Xiaodong Tan and Cristina Kountiou