The persistent-sensing requirement sits at the center of the AI wearables debate. For a face-worn device to deliver contextually useful artificial intelligence, it needs a continuous audio and visual feed from the wearer's environment. The architecture is not optional: inference quality scales directly with context, and context means running ambient sensors all day.
Big Tech has concluded that wearables, particularly those worn on the face, are the gateway layer for consumer AI distribution. The proximity argument holds up mechanically. The closer a device sits to the wearer's sensory environment, the richer the data stream the model can draw on. Face-worn hardware wins that contest.
The stalkerware label applied to the category reflects a different reading of the same architecture. Continuous ambient capture, by design, records bystanders who have not consented to be in anyone's inference pipeline. "Cringe," the other term critics have attached to the category, runs parallel: it marks the wearer as someone imposing a surveillance apparatus on the people around them rather than as an early adopter of useful technology.
The commercial tension is real and not easily engineered around. Big Tech's distribution thesis requires mass adoption. Mass adoption requires social permission to use the hardware in ordinary public settings. A platform that generates friction every time it appears in a coffee shop or a meeting has an addressable market considerably smaller than its technical ambitions.