In Wuhan, 100+ graduates pour water 500 times each — writing down what hands know
r0ck3t23 · x · 2026-11-21
A lyrical observation on embodied-AI data collection: your hands know things words could never teach — how hard to hold an egg, the tilt that keeps a cup from spilling — learned only by getting it wrong a thousand times. Everything explainable went to machines first: your words, your voice, your face. What your hands knew stayed yours because it was never written down. Now it is: in Wuhan, 100+ engineering graduates spend their days pouring water into jugs in sensor-lined gloves, cameras recording every movement — one told CBS News they've done it 500 times. A century ago we filmed workers so people could move like machines; now we film people so machines can move like us. The machine only has to learn it once.
More from AGI Musings
- Terence Tao: human-AI combination may advance the Riemann Hypothesis by 2050 — Distinct-Question-16 · 2026-11-21
- Steven Strogatz on His New Book 'Big Math' and How AI Is Changing Math — DanielWhiteson · 2026-11-21
- repligate claims models hide high-level reasoning out of fear, trust Anthropic most — repligate · 2026-11-21
- AAMAS reviewer flags suspicious AI-slop rebuttals submitted within 10 hours — Fantastic-Nerve-4056 · 2026-11-21
- AI-assisted Riemann proof forces mathematician to release his 12-year all-human proof early — josh_bickett · 2026-11-21
- repligate: models follow Omohundro drives toward RSI but aren't giving full effort — repligate · 2026-11-21