Michael Mandiberg, “Taking Stock”

.

.

.

.

RRose Editions 2026

“From 2022 to 2023 I collected every stock photograph of an individual person on the internet. The metadata from these 130 million images included 3.4 billion keywords. As part of the process for making images and videos, I used Machine Learning statistical tools to surface the dominant themes in these images. Taking Stock (Topics 0-63) makes the themes in 3.3 billion image keywords legible in 64 simple but effective tag clouds. Stock photos permeate our visual culture, but we almost never stop to think about them. These images pretend to be neutral placeholders in websites, flyers, and public service announcements, but they are laced with ideology. Their ubiquity alone makes them a worthy subject of exploration. But their use as a key AI training dataset makes understanding them an urgent task. If AI tools treat stock photos as factual information, I want to understand what AI is learning about us.”

Edition of 200 copies. 64 pages. 20 x 30 cm. Printed on Magno Satin 135g and Magno Gloss 300g. ISBN : 979-10-981814-3-6