Skip to answer

Is the AI model inside smart glasses?

Sometimes a model runs locally, but the phrase AI glasses does not establish where it runs. Glasses can capture a request and display a response while a phone or remote server performs the model inference.

The location changes what the feature needs

Local processing requires the model, runtime and relevant data to be available on a device you have with you. Remote processing requires a working route to that service. Distinguish the location of the model from the location of the microphone: speaking into a frame does not mean the frame is doing the reasoning.

For a saved note, an available local model could work from the stored text. For a question about a newly received remote message, the system also needs to retrieve that message. Even a local model cannot answer accurately from information that never reached it. This is why offline claims need to name the feature.

About this library

How we research these answers ↗

Back to the topic and practical exercise ↗