A decade after Google Glass crashed and burned in the public eye, Google is trying again with a new prototype that leverages its Gemini AI assistant. The company showed off the device at its Google I/O 2026 conference, and early impressions suggest that this time, the world might be ready.

The prototype, manufactured by Samsung, is not a final product but a demonstration vehicle for Google's vision of connected eyewear. It combines elements from partnerships with Warby Parker and Gentle Monster, aiming for a design that looks like normal sunglasses. The most important achievement is that the glasses feel ordinary — there's no extra weight, no awkward bulk, nothing to announce that you're wearing a computer on your face.

This is a stark contrast to the original Google Glass, which debuted in 2012 as Project Glass. That device was clunky, conspicuous, and quickly earned its wearers the derogatory label "Glassholes" because of the outward-facing camera that could record without clear consent. Society has since softened to smart glasses, thanks largely to Meta's Ray-Ban Stories and subsequent models. Even so, privacy concerns remain — reports of users disabling the recording LED on Meta glasses have kept the debate alive.

Google's new prototype includes a small camera notch on the right side of the frame, a button for snapping photos, and a touch-sensitive surface for swiping and tapping to activate Gemini or adjust volume. The integrated speakers are present but were faint in the noisy demo environment. The glasses connect to your phone for Wi-Fi or cellular data, essentially acting as an interface to Google's AI.

What can Gemini do in these glasses?

Gemini serves as a hands-free search engine. It can explain what ratatouille is, play songs from Spotify, add reminders, and look at the world through the camera to provide contextual information. For example, it can identify a concert poster and find tickets, or suggest wine pairings for a dinner recipe. All of this is delivered via the speakers, eliminating the need to pull out your phone.

That convenience is the core value proposition. Instead of fishing your phone out of your pocket, unlocking it, opening an app, and typing or snapping a photo, you simply tap the glasses and speak. The process is streamlined, but it's not revolutionary — it's an evolution of the voice assistant concept we've had for years on smartphones and smart speakers.

The heads-up display (HUD) is minimal. It shows reminders and basic navigation aids, but during the demo, the mapping feature displayed an overall map rather than turn-by-turn directions. Google says the final product will allow step-by-step guidance, which would be handy for navigating airports, campuses, or museums.

Some AI features feel forced. The ability to restyle a photo into an anime is unlikely to be a killer app. But the core utility — asking questions about the world around you — feels genuinely useful. For instance, while walking through the Shoreline Amphitheatre complex, Gemini could identify buildings and provide information.

Privacy and social acceptance

Google was careful to note that no facial recognition is being built into the glasses. The camera is activated only by a deliberate tap or by saying "Hey Google" (a wake word is planned for the final product). This overt signal helps address the privacy concerns that plagued the original Glass.

Still, the idea of having a conversation with someone and then interjecting a command to Gemini to set a reminder or ask a question is polarizing. Some will find it natural; others will find it intrusive. The glasses are designed to always be listening for a wake word, which is a double-edged sword for privacy advocates.

The prototype's battery life remains an unknown. Google did not disclose specs, and the demo was too short to test endurance. For a device that is meant to be worn all day, battery life will be a critical factor.

Historical context: From Glassholes to mainstream?

The original Google Glass was ahead of its time. It was launched in 2014 as a $1,500 Explorer Edition for developers, but the public quickly rejected it on privacy grounds. Cameras were banned in many venues, and wearers were asked to remove them. The project was eventually killed off in 2015 but live on in enterprise editions focused on manufacturing and medical fields.

Fast forward to 2026: smart glasses have slowly entered the mainstream. Meta's partnership with Ray-Ban has produced stylish frames that record video and take calls, and Amazon's Echo Frames offer audio-only assistant access. Google's approach combines both: a minimal HUD for visual information and a powerful AI assistant in the background.

But challenges remain. The HUD is basic, and the glass lenses are not prescription-adjustable yet. Google plans to offer lenses through Warby Parker and Gentle Monster at launch, but the prototype I tried used a single focal length for the display window, which might not align with everyone's vision.

The glasses also require a constant phone connection, which drains both devices' batteries. The long-term vision is to make the glasses standalone, but that would require significant advances in battery technology and miniaturization.

In many ways, the new glasses are what Google Glass always wanted to be: unobtrusive, helpful, and socially acceptable. The AI integration is the key differentiator, turning the glasses into a credible phone replacement for quick tasks. But the device is not a product yet — it's a prototype meant to show what's possible. The real test will come when Google announces a shipping date and pricing.

One area where the glasses could excel is in guided experiences. Imagine walking through a museum and having Gemini whisper the story behind each painting as you glance at it. Or navigating a conference hall with overlaid directions. The potential is there, but Google needs to refine the HUD and ensure the AI is accurate and responsive.

Ultimately, the success of Google's AI glasses will depend on how well they integrate into everyday life without demanding too much attention. The prototype shows promise, but the final product must solve the battery and privacy equations to achieve mainstream adoption.

Source: PCWorld News