Hands-On with Google AI Glasses: The Future's Almost Here

During Google’s I/O developer conference this week, we experienced a brief hands-on demonstration of the company's forthcoming AI-powered glasses. These were not the audio-only glasses slated for shipment this fall, but rather the more advanced version offering a combined audio and visual interface.

First unveiled at last year's event, these Android XR glasses feature an in-lens display that overlays valuable information onto the real world. This includes dynamic widgets capable of showing details such as weather forecasts, walking directions, Uber pickup specifics, live translation, and more – even personalized widgets designed by users with AI assistance.

Google confirmed that both the audio-only and the future display-equipped versions of the glasses will be compatible with both iOS and Android smartphones.

This display-enabled eyewear is envisioned as the evolutionary successor to the initial generation of audio-only glasses, expected later this year. The design of the glasses themselves is a collaborative effort with Warby Parker, Gentle Monster, and Samsung, integrating Google's technological advancements with the distinct aesthetic sensibilities of these brands.

The specific glasses we tested were unequivocally a prototype, albeit one refined enough for external evaluation. Representatives demonstrating the XR glasses explained that this prototype allowed Google to prioritize experimentation with the display technology and its impact on battery life, without being constrained by the cosmetic details of various styles and shapes. Consequently, these spectacles differ significantly in terms of fit, form, dimensions, and finishing from any future commercial release. Their purpose was akin to exploring the internal mechanics of the glasses within a fundamental, comfortable frame.

While the final shipping version of the glasses will incorporate a feature to detect when they are worn or removed, this functionality was absent from the prototype we engaged with.

Activating Gemini involved a two-second press on the right side of the glasses' frame, accompanied by a startup chime confirming Gemini was active and listening. In the demo version, initiating Gemini also simultaneously activated the camera. However, the commercial release will grant users the option to configure whether the camera turns on automatically with Gemini.

In an initial test, we requested Gemini to play music from a favorite artist. Unfortunately, the noisy venue made it challenging to properly assess sound quality; even at maximum volume, the music was difficult to hear with clarity and detail. This limited experience suggested the glasses might not replace high-quality earbuds, though they would be adequate for casual listening during outdoor activities like walking, hiking, or household chores. A notable advantage over earbuds is the enhanced ability to hear ambient conversations, surpassing the transparency modes found in devices such as Apple’s AirPods.

To cease music playback, a single tap on the middle side of the frame, similar to tapping one's temple, sufficed.

For the second test, we used a finger to press the photo capture button, taking an image of a person. With the display inactive, the photograph was transferred directly to our phone and watch. (Future iterations will support video capture via a long press, a feature unavailable on this prototype. Video capture would typically provide a thumbnail preview rather than a still image.)

Users will also be able to instruct Gemini to take a photo without pressing a physical button, and even apply AI manipulations to the result. For instance, one could command, "take a photo and turn the person into an anime character." The image would then be sent to the phone, processed by Gemini and Nano Banana servers, and returned in its edited form.

At the Google I/O venue, where Wi-Fi networks were under considerable strain, this entire round-trip process took approximately 45 seconds.

When the display was active, a straightforward home screen appeared within our field of view. The demo unit featured preloaded widgets displaying the weather and a countdown to the Google I/O event. Users could also configure quick launchers for frequently used applications like Google Maps or Translate, aligning with their primary use cases for the glasses.

The prototype was equipped with a single display over the right eye, though the underlying platform is designed to support both single and dual displays, as well as the audio-only variant. The image itself appeared slightly fuzzy, which we attributed to our bifocal prescription contact lenses (one lens for distance, one for near vision). Closing one eye improved focus, but the experience quickly induced some eye strain above the right eye, leaving the exact cause, including our prescription, uncertain.

One of the most impressive demonstrations was the real-time language translation capability, powered by the Google Translate app on a paired phone. As a demonstrator spoke rapidly in Spanish, the glasses automatically detected the language, displayed the English text translation on the in-lens screen, and simultaneously had Gemini speak the English translation into our ear. This feature alone could be a significant draw for global travelers.

It is worth noting that the Translate function will also operate on the audio-only glasses, albeit without the text appearing directly on the eyewear. Instead, users could view the transcription on their phone, if desired, in addition to receiving real-time audio feedback.

Another compelling demonstration involved using the glasses for navigation. While leaving the venue to test accuracy outdoors was not feasible, we gained a clear understanding of its functionality. Users could initiate Google Maps navigation by simply asking Gemini for directions to a destination, even if vaguely phrased, such as "the nearest coffee shop."

Gemini would then activate Google Maps on the paired phone, eliminating the need to physically retrieve the device. After a brief loading interval, the glasses would display turn-by-turn directions. When looking forward, the next turn's information was prominently shown. For spatial orientation, looking downwards revealed a "blue dot" on a map, allowing users to rotate their head left or right to align their position, much like orienting a map on a smartphone.

Subsequently, looking up again allowed for continued walking without the map obstructing the forward view.

Given the integration with Google Maps on the user's phone, pre-saved destinations such as "home" and "work" would be readily accessible.

We also briefly utilized the glasses to identify various objects in our field of vision and pose questions about them. The glasses initially struggled to identify a replica of a Monet painting on a nearby shelf. This was due to the prototype not automatically enabling the camera; it required manual activation via the app. Even after moving closer to focus on the Monet signature in the bottom left, it took several prompts before Gemini confirmed it appeared to be a Monet.

Other identification tests proved smoother, with the glasses promptly recognizing a plant on the shelf and answering queries about different recipes in a book. While acknowledging that these capabilities are currently available through Google Lens or other AI-integrated chatbot apps, the hands-free nature of performing these tasks without retrieving a phone presents an interesting advantage.

Google has indicated it will release further details about its Android XR display glasses later this year, coinciding with the expansion of its trusted tester program.

In the interim, the company suggests that audio-only functionality will sufficiently meet the needs of some users. This perspective could be seen as a strategic framing given that the display-equipped glasses are not yet fully ready, particularly in light of competition from companies like Meta and Snap in this sector.

Similar to the display version, the audio glasses also offer access to Google’s Gemini AI, with responses privately delivered through the frame's integrated speakers. Users can perform actions such as listening to music, capturing photos with a button press, making calls, or interacting with phone applications, mirroring the capabilities planned for the future display versions.

While the demonstration did not include interaction with third-party applications, the glasses are designed to allow users to instruct Gemini to perform tasks like, "take the ingredients from this recipe and add them to my shopping list."

In another example presented during the event's keynote, the glasses showcased their ability to visually analyze a meal being cooked on a stove and provide AI-driven feedback, such as whether the meat had reached its optimal doneness.