David Collingridge’s 1980 book The Social Control of Technology introduced what became known as the Collingridge dilemma: technologies are easiest to steer while they are still new and their consequences are hard to foresee. By the time the consequences become clear, the technology is often so embedded in daily life that changing course is nearly impossible.
We have already lived through one version of this. Social media promised connection and access to information at a scale no previous medium could match. Two decades later, we are still negotiating with its effects on attention, political discourse, and adolescent mental health — largely after the fact, because the habits and business models had already hardened before anyone had a clear view of the costs. Collingridge’s point wasn’t that we should have known better in advance. It’s that the window for shaping a technology closes quietly, well before its effects are visible.
AI glasses are a good test of whether we learn from that pattern.
We already worry about surveillance, privacy, identity theft, and hacking. But a technology built to be worn all day, whose purpose is to remove the friction between us and information, may create quieter forms of dependence that are harder to name.
Here is one possibility: we may become less curious.
The danger isn’t that we will learn less, but that we may eventually wonder less. The link curiosity-learning may become weaker and weaker.
Consider an urban-exploration course, where students are sent into a city to cultivate attention: to notice something unusual, follow a clue, ask a question, make a connection. Glasses that instantly name a building and recite its history could short-circuit the very experience the course exists to create. The technology tells us what to notice, instead of letting us discover what catches our own attention or follow a hunch.
This isn’t speculation. Research on the “Google effect” has shown that when people expect information to stay available online, they remember less of the content itself and more of where to find it (Sparrow et al. 2011). Related findings link habitual GPS use to measurable decline in spatial memory. The pattern is consistent: outsourcing a cognitive skill tends to atrophy it, even as it makes us faster in the moment. There’s no obvious reason why curiosity would be exempt.
A fair objection arises: couldn’t glasses increase curiosity, by surfacing facts we’d never have thought to ask about? Learning that the building in front of you sits on a fault line, or was the site of a forgotten strike, or that some famous lived there, might prompt more questions, not fewer. That’s possible – and it depends entirely on design, on whether the technology is built to prompt inquiry or foreclose it. But the default incentive for a product designed to feel effortless is to resolve the gap rather than open it. Friction is exactly what these products are built to remove, and friction is where curiosity tends to live.
If the gap between seeing and knowing disappears, so do some of the spaces where curiosity is born. Curiosity requires a bit of ignorance. Could this be the end of serendipitous discovery – in travel, or in a walk through a familiar neighborhood? We are already seeing some of this with various recommendations as to where to eat or what to see.
There is no AI substitute for curiosity.
Collingridge’s lesson is that the time to think about how we want AI glasses to shape us is before they become indispensable, not after. Concretely, that means building in friction on purpose: an interface that offers a question before an answer, a delay before disclosure, a mode that can be switched off without switching off the product. None of that happens by default. It has to be designed in while the technology is still young enough to be argued with.
This is also a small case study in a much larger pattern: revolutions tend to overshoot by borrowing the vocabulary of one thing to sell another. “Intelligence” implies augmentation; what’s often being sold is substitution. Getting that distinction right – for AI glasses, as for AI more broadly — is most of the argument.
—
Sparrow, B, J. Liu, D.M. Wegner, Google Effects on Memory: Cognitive Consequences of Having Information at Our Fingertips, Science, 333, 776-778 (2011).
