Machine Hallucinations
summary
A studio builds immersive data artworks that train 100+ AI models on public archives and use real inputs—113 million images for a New York installation, NASA space data, live wind feeds and a 100 billion–point Gaudí façade—to generate dynamic, nonpersonal sculptures. Outputs include an 18‑channel projection with 32‑channel sound shown to ~300,000 people, an NFT room scented from a model trained on half a million molecular profiles, a MoMA piece using nvidia salgent ii ada algorithm, and public works reaching 1.7M and 200k viewers.
AI-written summary · about
chapters (4)
related talks
26mInfinite AI
38mMaking Art with Artificial Intelligence: Artists in Conversation
43mNew Age Of Renaissance: How to Integrate Art, Science, Technology & AI into Public Spaces
41mArt and Technology Experiments with Google Arts and Culture
15mGenerative Design
15mMuseums Beyond NOW
47mExperiential and Immersive Digital Art
44mFrom On-site Exhibitions to Digital Art, a Journey Through Different Dimensions