I’ve Always Been Jealous of Other People’s Families is a prologue to Marrow. A project that explore the possibility of mental disorder in the intelligent machines we create.

This interactive installation embodies ML processes, offering insight into the inner world of GAN (generative adversarial network). 

Four people gather around a dinner table, acting as family. The table is set with sound, projection, and text, eliciting something familiar yet disturbing. Each “family” is processed through the logic of GAN via different machine learning networks which are trained on a dataset of “Perfect Family Dinners,” and sees participants through that lens. The output projected is a distortion, revealing an attempt to fill in the gaps between fantasy and reality. Looking into these outputs we found hidden pockets of error and loss.

Can this be the portal to explore machine’s mental state?
One aspect of mental illness is that reality is experienced through a specific lens, not necessarily related to the input received. “I’ve always been jealous of other people’s families” is a prologue to Marrow, a project which explores the possibility that the intelligent machines we create can develop a mental disorder.
︎ Visit project’s website 

My role: concept and creator.
Additional credits ♡ Cristobal Valenzuela (Technical Director), Laura Juo-Hsin Chen 陳若昕 (Animation and developer), Avner Peled (Speech and sound design), Ziv Schneider (Advisor).
Executive producers: Arnaud Colinart and Hugues Sweeney. Producers: Emma Dessau and Louis- Richard Tremblay.
Production of Atlas V, Raycaster and National Film Board of Canada.

Date: November 2018
Platform: ML interactive installation
Technology: Denspose network, pix2pix network, AttnGan network, Supercolider.Featured: IDFA DocLab 2018. Supported by: IDFA DocLab and MIT Open Documentary Lab.
More on the research: Family2Family/ published at Immerse. When Machines look for order in Chaos/ published at Immerse. 5 reasons why I believe AI can have mental illness/ published at becoming human.

︎   ︎