Hito Steyerl, Mean Images, NLR 140/141, March–June 2023

An artist considers a new form of machinic representation: the statistical rendering of large datasets, indexed to the probable rather than the real of photography; to the uncanny composite rather than the abstraction of the graph.

https://doi.org/10.64590/uhm 140/141 , Mar June 2023 Hito Steyerl Mean Images Share A while ago, science-fiction writer Ted Chiang described Chat gpt ’s text output as a ‘blurry jpeg of all the text in the web’—or: as a semantic ‘poor image’. footnote 1 But the blurry output generated by machine-learning networks has an additional historical dimension: statistics. Visuals created by ml tools are statistical renderings, rather than images of actually existing objects. They shift the focus from photographic indexicality to stochastic discrimination. They no longer refer to facticity, let alone t

related reading

• Lonely Surfaces: On AI-generated Imagestheconvivialsociety.substack.com
• The Illustrated Stable Diffusion – Jay Alammar – Visualizing machine learning one concept at a time.jalammar.github.io
• The Dark Forest and Generative AImaggieappleton.com
• The Platonic Representation Hypothesisphillipi.github.io
• Epistemic calibration and searching the space of truththesephist.com
• Machina Lucida: Reflections on Artificial Images by Lucas Gelfondspikeartmagazine.com
• Large Language Model: world models or surface statistics?thegradient.pub
• ⭐️ Diffusion Modelsandrewkchan.dev
• floatingprojectscollective.net/art-notes/mining-datafied-worldfloatingprojectscollective.net
• [2210.13382] Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskarxiv.org
• [2006.11239] Denoising Diffusion Probabilistic Modelsarxiv.org
• https://arxiv.org/pdf/2006.11239arxiv.org