Learn AI · Level 3 · Tech and tools · step 15 of 17 · word 3 of 6
What is edge AI?
Edge AI means running AI on the device itself, such as a phone, car, camera or watch, instead of in a data centre. Federated learning trains models across devices without collecting the data.
In plain English
“The edge” is where the data is created: your phone, a factory sensor, a car. Running AI there is faster, works without internet and keeps your data on the device. Phones now include special chips called NPUs for this, and models are shrunk with tricks such as quantization.
An example
In 2017, Google started testing federated learning in Gboard, its Android keyboard. Each phone learns from how its owner uses suggestions and sends back only a small model update, not what was typed. The updates are averaged into a better model for everyone.
Watch out
Devices have little memory and battery, so edge models are smaller and less capable than the big ones in the cloud. Many products mix both: quick, private tasks on the device, harder ones sent to a data centre.
Learn AI path · Level 3 · Tech and tools · step 15 of 17
Next wordHBM (high-bandwidth memory) →See all of level 3