In the wake of breaking the news of Meta’s proposed “Name Tag” feature, the New York Times ran another, seemingly more anodyne, story on the smartglasses. The piece covered the emerging trend of dining influencers wearing the glasses to record their meals and create content, and the comments section, predictably, went a bit wild, with calls for restaurants to ban the glasses and threats of violence against people wearing them. Not surprisingly, there were very few comments that also suggested banning phones or security cameras from the same establishments.
That same week, I personally made a rookie mistake that I’ve managed to avoid thus far – trusting information an LLM served me as a fact without double checking. I was drafting a LinkedIn post, and needed some quick data, and the LLM gave me outdated information. In the end, it wasn’t a huge deal, but I was certainly embarrassed when someone in the comments corrected me. I vowed to always double check and start using another LLM for research.
These two seemingly disparate events point to something bigger, though – and a problem that could doom adoption for both emerging technologies. We are mired in a moment of low trust in institutions already – and now, that lack of trust has started to spill over into our relationships with platforms and devices. And if those companies don’t work to repair the trust, things could go south very quickly.
“As a former CEO and tech founder, I’ve spent my career making decisions with incomplete information, so uncertainty never scared me,” says Anat Baron, a futurist and AI keynote speaker. “What scared me was noticing I couldn’t always tell where my judgment ended and the machine’s confidence began. Over a few months of working with LLMs every day, I noticed I was surrounded by systems that sounded certain, persuasive, polished. And often wrong. The unsettling part wasn’t that I thought that AI will replace me, it was abdication. That quiet moment when you stop arguing with the machine because the answer sounds smarter than you do. I caught myself double-checking my instincts more than I was double-checking the outputs. For me, that was a flashing red light. Ultimately, the bigger risk isn’t replacement, it’s abdication of responsibility.”
Part of the reason so many people choose to simply believe an LLM when it serves them information is because it sounds so confident. Baron calls this “authority drift” – the moment when AI starts shaping decisions not because it has earned authority, but because it projects certainty.
With AI, it’s so confident and it’s so clear, and it stands up for itself that you are going to believe it – until you don’t,” says Baron. “That’s why it’s so seductive.”
While most LLMs are generally contained within apps or called out in search results, in future they will become the plumbing of the internet, and that’s a big part of what could make smartglasses so valuable. The positive future looks like this: real-time, contextual information delivered right to your eyeballs. If it works out, it could be revolutionary, but in an era of already low trust, will people make the leap?
In order to have trust, you need named human owners monitoring the inputs and the guardrails,” says Baron. “Every meaningful decision still needs a human owner, even when AI is in the loop.
So what can be done to start repairing trust and making sure that these technologies benefit us in the long run? The first step is to acknowledge that, just like humans, LLMs will make mistakes – and design workflows where humans are responsible for catching and correcting those errors whenever possible. Making a habit of always checking primary sources was useful before the days of widespread AI adoption, and the Reaganite notion of “trust, but verify” seems poised to make a comeback.
The other solution is to make sure that people are, in most cases, transparent about how and when they are using their smartglasses to make content. Making the recording lights brighter, making it a point to ask permission, and disclosing when recording a person one-on-one are best practices that, while they probably can’t be legislated, could be adopted as norms via public service campaigns. The internet’s boyfriend Hudson Williams demonstrated best practices for this at a recent party, and just like telling your friends you’re going to take a photo with your phone, telling them you’re going to use your glasses should become a norm.
Baron’s “Human + AI Equation” framework is all about assigning clear decision rights – what stays human‑led, what gets AI augmented and what can be automated. She argues that the same logic needs to apply to smartglasses and other emerging interfaces, or trust in the whole ecosystem will erode.
