Royal College of Art
MA Information Experience Design
Created withProcessing / OSC / Neural-storyteller / MS COCO captions
As machine learning sits on the threshold of glitch-free interpretation and calculation of the world around it, we lose comprehension of the inner workings, and fallibilities become hidden within perceived accuracy. This moment, on the precipice, we can still see ourselves outside of computation, but whether this separation lasts or whether society becomes absorbed into an objectivity defined by computational machines remains to be seen.
Every 30 seconds a camera captures the audience’s image. By analysing the objects in the photo, the machine attempts to match what it sees with the most appropriate caption from a pre-written set. Projections of nearest and tenth closest captions shown on the adjacent walls.
As part of a larger body of research into the nature of objectivity, this installation aims to demonstrate both the ability and the limits of image classification as a critique of the influence of a training sets outputs.