Member-only story
iFlytek’s WAIC 2026 AI Blackboard: A Technical Breakdown
A 92 second clip from WAIC 2026 has been sitting in my head for three days now.
A presenter stands at a plain looking blackboard in Shanghai. He writes y = x² — 3 in plain text, no different from what any math teacher has written for the last hundred years. Within a second, a clean parabola opens up next to his handwriting on the board itself. He then sketches a rough triangle, freehand, the kind of wobbly shape any of us would draw in a hurry. The board straightens it into a proper triangle, tags the geometry concept it belongs to, and then lifts the flat drawing into a rotatable 3D solid that students can spin around to look at its edges and faces.
I have spent the better part of this year writing about GPU inference, KV caching, and how AKS clusters serve large models at scale. So my first reaction watching this wasn’t “wow, magic.” It was “okay, but what is the request path here.” What has to happen between chalk touching slate and a 3D model appearing, and how fast does that round trip actually need to be for a classroom of sixteen year olds to not get bored waiting.
