Centaur Drawings
summary
An exhibition recreates algorithmic co‑creation by pairing a pen plotter drawing robot with human visitors and a homebuilt LSTM in Python that generates asemic characters. The team collects 31,300+ original hand drawings as a training set, trains a small recurrent model, and produces ~30,000 algorithmic variants; outputs appear as automated chess maps, flow‑field diagrams and interactive live drawings. Code and hardware are open on site, visitors draw with the robot, and the pieces successfully visualize human labor and system aesthetics.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:50Exhibition concept
- 3:14Kasparov’s Centaur Chess
- 6:26Automated chess maps
- 9:58Neural networks generating drawings
- 13:52AI model process
- 16:46Visitors interpreting symbols
- 19:33Systems theory and aesthetics
- 21:39Public interaction with robot
- 23:25Collective labor and AI
- 25:23Q&A
related talks
38mMaking Art with Artificial Intelligence: Artists in Conversation
12mAlgorithmically Generated Visual Designs
26mInfinite AI
06mCan Children Emulate a Robotic Non-Player Character’s Figural Creativity?
39mGenerative Machines
32mCode goes in, Art comes out
25mTurning Data Into Generative Chaos
46mThe History of Computer Art