The first serious attempt at consumer smart glasses failed for reasons that had almost nothing to do with the hardware. The devices worked adequately. People did not want to be near someone wearing one.
The lesson that was learned
Look like glasses. The current generation is built on ordinary frames from established eyewear brands, and the design goal is that nobody notices. That is a substantial improvement over hardware that announced itself from across a room.
The capability set was also narrowed deliberately. Audio, a camera, a voice assistant, and in some models a small display. Not an attempt to put a computer on your face.
What actually gets used
Audio, mostly. Open-ear speakers that leave you aware of your surroundings turn out to be genuinely useful for navigation, calls and listening while doing something else.
Photo and video capture from eye level is the second thing, and it is where the social problem persists. An indicator light is a weak signal, and everyone knows it.
The display problem
Projecting a bright, readable image into the eye through a transparent lens, in daylight, without adding weight or heat, remains genuinely difficult. Current displays are small, dim relative to sunlight, and limited in field of view.
They work for glanceable information – a direction, a notification, a translation. They do not work for anything you would want to read.
Where the AI angle helps
The most compelling demonstrated use is contextual: ask about what you are looking at and get an answer. Translation of a sign, identification of an object, a reminder tied to a place.
That works because it does not require a display at all. Audio answers a visual question, which sidesteps the hardest engineering problem entirely.
The prediction
These succeed as audio devices with cameras and assistants, sold as eyewear. The augmented reality version – persistent, rich, spatially anchored graphics – remains a different product that current optics cannot deliver at a size anyone will wear.