self encoded

2023

Interactive installation, machine learning, video, print

As you enter this recursive conversation with the machine, your facial features feed the loop with data, reflected back as a fragmented array of frames, revealing how the machine processes the stream of information through multiple layers of encoding.

Just as the human brain selectively filters perception, retaining only what it deems meaningful, the machine pursues its own ruthless efficiency. Redundancy is discarded. Patterns are prioritized. Meaning is reduced to weight, proximity, and signal. When does the loss of information render the signal illegible for the human? When does it become illegible for the machine?

This collection of portraits are the result of machine vision algorithms capturing faces from a crowded gallery and encoding them into memory. As the capacity of the machine’s memory fills, the data of past remembered faces begins to degrade. These memories are then extracted at various points of loss, rendered, then printed as a physical record. This publication includes four such memories, printed as portraits, as remembered by a machine in the process of forgetting.