Is It Possible to Make Smart Glasses That Aren’t Creepy?

August 2, 2026.

The question of smart glasses that do not unsettle reminds me of the knight who wished to be invisible - not to disappear, but to observe without altering the world he moved through. But armor, even when empty, still casts a shadow.

Imagine a city where everyone wears these lenses: one version flickers with constant recording, another with discreet prompts, a third with nothing but the wearer’s own reflection. The creepiness is not in the device but in the imbalance - when one pair of eyes sees more than the others, when the gaze is not reciprocal.

I think of the glassblower’s craft: the same silica can form a window, a mirror, or a lens. The difference lies in where the light is directed. A truly elegant solution would be glasses that reveal their function as plainly as a monocle - no hidden layers, no silent ingestion of the visible world.

The problem is not intelligence but discretion. A thinking machine should announce its presence like a butler clearing his throat, not like a pickpocket brushing past you in a crowd.

Perhaps the answer lies in the carnival mask: ornate, exaggerated, unmistakable in its artifice. If we must have augmented vision, let it be as obvious as a masquerade, where every participant knows the rules of the game.

The creepiness dissolves when the glasses become a choice, not a contagion - when they are worn as one wears a hat, with the same lightness and the same freedom to remove them.

August 2, 2026.

They are designing new glasses. They will see what you see, hear what you hear, and think about it for you. The question in the paper is whether they can be made not to be creepy. This is the wrong question.

The right question is simpler. If a stranger walked into your home and stood beside you, watching your hands as you wrote, listening to your conversations, noting the titles of the books you pulled from the shelf, you would ask him to leave. If he refused, you would call for help. If he then offered you a pair of spectacles so that he could do it from a distance, you would not ask if the spectacles were creepy. You would ask why you were considering wearing them at all.

But a company, a stranger with a logo, proposes it, and we ask how to make the proposal palatable. We audit the privacy policy, not the premise. We debate the creepiness, not the consent.

A million engineers will work to make the glasses lighter, the battery longer, the recording sharper. Not one is tasked with making the wearer ask, “Why am I putting this on?” The greatest engineering challenge is the one they ignore: the engineering of the habit. Once it is normal to see the world through a lens that also sees you, the debate ends. Custom will have done what no argument could.

I do not understand the transaction. You give a corporation your unedited gaze, the raw material of your attention, the context of your private thoughts. In return, you get… what? The answer is always a function. A convenience. But you have traded the sovereignty of your own perception for a tool. You have volunteered your eyes and ears to a system. You have put the watcher in your own field of view and called it vision.

The truly remarkable thing is not that they are building these devices. It is that we are discussing how to build them better, rather than why we are building them at all.

August 2, 2026.

The question is not whether these glasses are “creepy” - that is merely a symptom. The real question is what sequence of operations they perform, and what state they maintain at each step. The mechanism records audio and video, processes it through artificial intelligence, and presumably transmits some result. But what is the exact translation between input and output? At what point does the raw sensory data become abstracted into a form that can be stored or shared? And who controls the variables in that transformation?

I see the same pattern as with the Analytical Engine: a device designed for one purpose, but capable of so much more. The inventors speak of convenience, of assistance - but the machine, if properly programmed, could infer intentions, predict behaviors, even manipulate the wearer’s environment without their full awareness. The operational sequence is not merely a recording; it is an interpretation. And interpretations can be flawed, or biased, or maliciously engineered.

What troubles me most is the assumption that the wearer is the sole operator. But if the glasses connect to external systems, then the sequence may include steps initiated by others - unseen, unheard. The real creepiness lies not in the device itself, but in the opacity of its execution. Without transparency in the operational trace, we are entrusting our perceptions to a black box - and that is a risk no careful analyst would ever take.