A decade after Google launched Google Glass to spectacular failure, the company is trying again with a new prototype that leverages its online AI interpreter, Gemini. I had the chance to try out a prototype version of Google’s XR glasses at the Google I/O conference. The world—and I—appear to be more receptive to what Gemini can do when it’s plugged into your ear.

Google’s first connected glasses, Google Glass, debuted in 2012 as Project Glass. The new prototype, manufactured by Samsung, incorporates elements of technologies from Warby Parker and Gentle Monster. It’s not a product per se, but a vehicle to show off Google’s vision for connected eyewear. Honestly, the whole thing still has an aura of unreality about it—the new glasses don’t feel that different from what the company showed off at I/O 2026, more than a decade later. What has changed is public attitude: Google Glass wearers were once derided as “Glassholes,” especially those who recorded people using the outward-facing camera. Now, connected devices from Meta have been somewhat bafflingly accepted by society, even with anecdotal reports of users still taking pictures and recording video after modifying the eyewear to disable a small LED that signals recording.

Google’s AI glasses are surprisingly light

Google didn’t answer questions about the hardware specs, battery life, or features of the display and camera. But the most important thing is that they felt normal. There was no additional weight or thickness—nothing to indicate you were wearing anything other than a pair of sunglasses out in the bright California sun. From the front, a small circle signals the presence of a camera, along with a small button on the right frame for snapping a photo. That right-hand side also includes a touch-sensitive surface for swiping and tapping to trigger Gemini and control the volume of the small integrated speakers (which were a little faint for the busy demo room). These design choices are a plus: most technology today seems intent on reminding you that AI exists, but here, Gemini sits in the background and shows up only when needed.

What can Gemini do in a pair of connected glasses?

Gemini fulfills its traditional role as a search engine—helping explain what ratatouille is, for example, or how it’s made. You can play songs from Spotify or add a reminder to take the cat to the vet. But the glasses also allow Gemini an eye into the world: looking at a concert poster and checking ticket availability, or explaining what wine might pair with a dish you’re preparing. All of this is piped into your eardrums via the speakers. The glasses don’t stand alone; they connect to your phone, which in turn connects to Wi-Fi or cellular data. Convenience is the key selling point—eliminating the steps of fishing out your phone, unlocking it, opening an app, and so on. There’s no facial recognition being built in, I was told.

Some of the AI feels decidedly forced. For example, snapping a photo and then restyling it into an anime? Absolutely not. But the glasses can do it, and even send the result to your phone or connected smartwatch. The on-glass heads-up display (HUD) is simplistic, but that’s fine. Most people will likely find it most useful for popping up reminders or basic navigation aid. The mapping feature I tried showed an overall map of surroundings rather than step-by-step directions like “turn left in 100 yards.” I think I’d want that, especially when navigating something like an airport, college campus, or museum, as well as just looking for a flat white.

Hardware and privacy concerns

Battery life will be a major concern. I’m not sure what I’d want out of a HUD—probably minimal distractions, certainly. So far, Gemini must be activated by tapping and holding the eyewear’s button, though a wake word (“Hey Google”) is intended for the final product. Privacy remains a qualm: though an overt signal (a tap or wake word) is helpful, I’m still hesitant about holding a close conversation with someone and then interjecting Gemini to add a reminder or appointment. That is going to be a polarizing feature. In certain situations, I might actually like an always-on role—for example, audio guides at museums that steer you through exhibits, but imagine an AI popping up to relate a story about a painting after noticing you glancing at it. There’s room for exploration.

As a general aid, Google’s Gemini glasses do feel useful. The real test will be how the product evolves from prototype to consumer device. Meta’s Ray-Ban Stories and their successors have already proven that smart glasses can find a market, but the addition of AI vision and a HUD could differentiate Google’s offering. The company must also address the stigma of “Glassholes” by ensuring robust privacy controls—like a visible recording LED—and dispelling fears of constant surveillance. With the right balance of utility, discretion, and transparency, these AI glasses could finally realize the promise that Google Glass sought to deliver more than a decade ago.

Source: PCWorld News