The lesson
Ordering through a wall of noise
It is 12 noon at the hawker centre and the noise is from people talking and laughing, and woks clanging. A regular at the front of the queue places his order. Half his words get swallowed by the background noise. I catch it anyway: 1 Hokkien Mee, extra sambal. Between stirs, I glance at the picture of the PayNow tick on his handphone. In 1 second, I know the payment is good. Now hand those two jobs to a machine.
A picture is a grid of numbers
Start with the picture. To a machine, a picture is not a picture. It is a grid of tiny coloured tiles, millions of them. Each tile is just numbers for brightness and colour. Out of all that, the machine must arrive at the word Successful and a green circle with a tick.
The stack from the last lesson does the job. The first layer answers tiny questions about tiles. Bright here, dark there. Those answers stack into edges. The next layer asks tiny questions about the edges, and curves appear. Layer after layer, edges build shapes. Shapes build parts. A green circle. A tick. The word Successful. And the stack says: payment successful.
Listening is the same trick, applied to sound. A shout reaches a microphone as 1 long wave. The machine slices it into strips. Early units answer tiny questions about each strip. Higher up, strips stack into sound units. The units stack into words. The words, weighed against everything the machine knows about ordering noodles, become meaning. The roar of the hawker centre is still in there. It is simply outvoted, layer after layer, by the pattern that keeps saying extra sambal.
Your face, in tiles of you
This is already happening in ordinary life. Your phone unlocks by looking at your face. Tiles of you stack, until the stack says you. Say hey to your phone and it answers, turning strips of your voice into words. Some supermarket checkouts read a bun or a banana straight off the weighing machine, no barcode. It knows your face. It knows your voice. It even knows your groceries.
Notice it unlock
Your phone will unlock by looking at your face. Let it. Notice the stack that just ran, its tiny questions ending in the word you. Later still, say hey to your phone and ask for tomorrow’s weather. Hear noise become an answer. That is this lesson happening in your hand.