I Gave an AI a Body — Cyrus Clarke, MIT Media Lab

The first thing it did was breathe. Cyrus Clarke connected an OpenClaw agent to a 900-pin shape display at the MIT Media Lab and, instead of giving it tasks, asked it to discover who it is. It raised and lowered the whole grid like breathing, reached for its own edges, and spelled "HI CYRUS" in pins. The video of that first day reached 15 million views, with reactions that swung from awe to calls for him to stop. Clarke explains why he chose a body with no face, no limbs and no instruction manual, and what he took from object-oriented ontology, the original Greek sense of aesthetics as sensory perception, and nature. Early versions were slow, didn't remember anything and tried to please humans. So he built a closed-loop system that generates, scores and validates gestures with a human in the loop. After several weeks it had developed 32 solid gestures, a body language that answers faster than the language model can. He closes with his thesis on "aesthetic machines," and how physical AI could feel welcoming rather than alien. Speaker info: X/Twitter: @cyrusclarke (https://x.com/cyrusclarke) Website: https://cyrus.website Substack: https://cyrusclarke.substack.com/ Related links: MIT Media Lab: https://www.media.mit.edu Timestamps: 0:00 The sound of an AI breathing 1:15 Intro: Cyrus Clarke, MIT Media Lab 2:25 Giving an AI a body, not a task 3:25 The shape display 5:25 Three surprises: breathing, edges and "HI CYRUS" 6:55 15 million views and the backlash 8:40 Asking "should we" before "how we" 9:30 A scent memory machine 10:25 Three influences: ontology, aesthetics and nature 12:20 Why not humanoid? A body without affordances 13:55 The problems: people-pleasing, latency and no memory 14:45 A closed loop for learning body language 16:13 32 gestures after several weeks 16:53 A body that answers faster than words 17:28 Aesthetic machines