desktopscope.comWe may receive a commission on purchases made from links. Tablets are slowly growing out of the awkward...
Google’s Gemini Live Can Now See What You See, But How Smart Is The AI Video Feature?
Gemini Live with camera and screen share is a clumsy name for a shocking smart feature that can use your phone’s cameras or screen sharing functionality to see the world around you and answer questions in real time. The feature was first rolled out in early April to users who had either a Pixel 9 or Galaxy S25 phone, or who were subscribed to Gemini Advanced. Barely a week later — before many of those users had a chance to discover the feature — Google rolled it out to all Android users with devices capable of supporting it. The wide release demonstrates a level of confidence in the product, and, I have to say, it’s warranted.
As one of SlashGear’s resident AI experts, I’ve spent a lot of time using generative AI products. Among other things, I tested Gemini against ChatGPT and pitted the latter chatbot against Microsoft’s Copilot derivative. Most of the time, trying to find a real use for these products is an exercise in frustration and futility. However, after spending about a week using Gemini Live with Video in my day-to-day life, I am frankly in awe. This is the advanced smartphone assistant we’ve dreamed of since Apple first introduced Siri over thirteen years ago.
That’s not to say it’s perfect. Many of the same issues persist, including AI hallucinations and confident liar syndrome. So, let’s break down exactly what Gemini Live with Video can do well, what it struggles with, and why I still have some hesitations around using it despite my positive experience.
Gemini Live with camera sharing is a game-changer
Let’s start with what Gemini’s camera sharing feature does well, which is quite a lot. After the feature rolled out to my Samsung Galaxy S25 Ultra, I started by asking it to identify objects in my bathroom, which it did a decent job of, though not a perfect one.
When I asked it to center my icons in the Windows 11 taskbar, the results were mixed. It got me to the taskbar settings and had me switch the icons to the center, but it also told me to turn on the auto-hide feature.
Stopping at Barnes & Noble, I pointed Gemini Live at books I was interested in, and asked it to tell me about them. Gemini provided me with brief synopses, as well as direct review blurbs from prominent critics at The New York Times, The Guardian, and more. It’s worth mentioning that those critics almost certainly did not give Google permission to train AI on their work.
While gaming, I pointed my camera at my screen and asked for help with a quest. Gemini gave me precise and accurate information that helped me make the right choices. Again, it’s worth noting that Gemini seemed to source its answers from game guides by writers much like myself. I couldn’t help but think I’d be out of a job if everyone asked Gemini for tech help rather than reading SlashGear.
Sharing my screen with Gemini, I showed it some old family photos of my great grandparents as they prepared to flee Nazi Germany. It offered details about the locations the photos were taken in, the clothes people wore in them, and more. It was a genuinely meaningful experience that enhanced my understanding of my family history.
Gemini’s video capabilities still have some pain points
There were several times during my testing when Gemini Live with Video failed me. First, it consistently got caught in loops, only able to offer limited information. When I asked for the aforementioned reviews, it told me how funny one book was. But when pressed for more detail, it turned into a thesaurus, telling me the book had been “praised for its wit and humor” while failing to tell me what, exactly, reviewers had found so funny. This was a consistent theme in my testing. Similar loops occurred when I asked for granular details about most things.
It’s hard to overstate how impressive Gemini Live with Video is, despite its shortcomings. Apple is currently fighting a five alarm fire as its promised AI features languish in development, while Google is pushing out game-changing features on a random weekday. I can’t recommend it for anything too serious – a bad book recommendation wouldn’t have hurt anyone, but you certainly shouldn’t trust Gemini for legal or medical advice. There are also privacy risks that might make you think twice before pointing the Gemini camera at sensitive information. Still, for the most common, low stakes tasks, it’s a remarkable addition to your Android phone’s smart capabilities.
Source: http://www.slashgear.com/1838442/google-gemini-live-ai-video-feature-how-smart-explained/
You may also like
