Google tests AI glasses at I/O

Google’s latest smart-glasses push is now split between a product due this fall and a prototype that still looks unfinished. At Google I/O this week, the company showed reporters Android XR glasses that can answer questions about what the wearer is seeing, translate speech and text, and surface directions in a small in-lens display. TechCrunch, which tested the prototype on May 22, said the system could identify objects, describe scenes and offer cooking feedback, but also made mistakes and ran into battery constraints. Google said separately on May 19 that its first commercial version will be audio-only glasses made with Samsung, Gentle Monster and Warby Parker. ### Which glasses did Google actually show at I/O? Google used I/O 2026 to talk about two categories of “intelligent eyewear.” In a company blog post by Shahram Izadi, Google said audio glasses that deliver spoken help are launching first later this fall, while display glasses would show information in the lens when needed. TechCrunch’s hands-on covered the second category: prototype Android XR glasses with an in-lens display. (techcrunch.com) The outlet said the glasses overlaid Gemini-powered information onto the real world, including translation, navigation and contextual answers about objects in front of the wearer. ### What could the prototype do in a real demo? TechCrunch reported on May 22 that the prototype handled several practical tasks during demonstrations, including identifying surroundings, naming objects and offering suggestions based on context. (blog.google) One test involved cooking, where the glasses gave feedback on food preparation. Google’s own description of the broader glasses roadmap matched much of that feature list. (techcrunch.com) The company said users would be able to ask Gemini about what they see, get natural turn-by-turn directions, send texts, snap photos, translate speech and writing, and ask Gemini to complete multi-step tasks. ### Where did the demo fall short? TechCrunch said the prototype was not consistently accurate. The outlet reported that scene descriptions and object recognition sometimes faltered, and that battery life remained a visible limitation during testing. (techcrunch.com) PCWorld, which also tried a prototype at I/O, described similar limits from another angle. The publication said the glasses were lightweight and more socially acceptable than the original Google Glass, but called the heads-up display basic and cited battery concerns. (blog.google) PCWorld also said the device it tried was “not a product, per se,” but a vehicle to show what Google is aiming for. ### What is Google promising for the first commercial version? (techcrunch.com) Google said the first release will be audio glasses rather than the full display model shown in some demos. In its May 19 post, the company said those glasses will pair with phones, use Gemini through voice or frame taps, and support tasks such as directions, messages, calls, music, photos and translation. TechCrunch reported on May 19 that Google announced partnerships with Warby Parker and Gentle Monster to produce the new AI-powered glasses, and said the devices were designed in collaboration with Samsung and would work with Android and iOS devices. (pcworld.com) ### Who is building the glasses, and when are they coming? Samsung, Gentle Monster and Warby Parker are the named partners in Google’s current eyewear plan. (blog.google) Google said the first designs shown at I/O would launch as part of Gentle Monster and Warby Parker collections later this year. The next concrete milestone is the fall launch of the audio-only glasses. Google has not given a ship date for the display-equipped prototype that TechCrunch tested on May 22, and the company’s I/O materials described that category only as part of the broader Android XR roadmap. (techcrunch.com) (blog.google)