Welcome to WIRED’s Google I/O live blog. Several WIRED folks are on the ground in Mountain View, California: Boone Ashworth, Julian Chokkattu, Lauren Goode, Steven Levy, and Reece Rogers. Follow our live updates below.All the Big Announcements in One PlaceWe're going to head out and see some demos. Thank you for following along with our live blog!Demis name drops “the singularity” as coming in the future to end the day! For my more immediate future, the keynote is over and I'm rushing to the nearest restroom.Well, that felt like an abrupt end!11 hours agoLineOne theme of I/O has been how AI dramatically speeds up the rate of coding, creation, scientific discovery, etc. etc. But at almost 2 hours, they haven't cracked the code to speeding up keynotes.Ah yes, Google's annual “good guy” section where they aim to show you how they do indeed care about climate and humanity and all that. Last year it was Google's wildfire detection satellites. This year it is the goal of “solving all disease.”I asked Gemini what the longest Google I/O keynote was to date, and it was in 2013, which ran 3 hours and 5 minutes. Sorry to anyone who had to sit through that. We're coming up on two hours now.Demis going on about AGI again. One thing he's talking about is Gemini for Science. Interesting contrast with OpenAI which disbanded its science team.Remember, they’re not Smart Glasses, they are “Intelligent Eyewear.”Nano Banana on smart glasses is actually bananas. The demo worked!Nothing like doctoring photos in real time.When you trigger Gemini on smart glasses, it defaults to Gemini Live for that conversational back-and-forth experience. All of these smart glasses have cameras, because that's how Gemini sees the world.Sergey wanted to be here, but he's busy parachuting into the White House.14 years (!) after Sergey parachuted into I/0 to introduce Google Glass, we get a reference design for Google smart glasses.Here's the reference design from Google and Samsung for its smart glasses. Audio-only glasses will land first later this year, then versions with displays will come next year.Photograph: Julian Chokkattu/WIREDNow here's what the final versions will look like from partners like Warby Parker and Gentle Monster this fall.Courtesy of GoogleCourtesy of GoogleThis outdoor venue has more bugs than a Mythos eval.Flow gets an update. It can turn one single photo into 16 related AI-generated images. It can also turn simple audio sketches of riffs or melodies into slick, highly produced songs. In other words, an on-demand slop factory. Available today.Boone is right! There are too many bugs at the theater. Need to get back to a reporter's natural habitat: inside. “Gemini, please zap those flies."Of all these changes to Flow, Google's AI creation software, the “avatar” feature, where users can essentially create a deepfake selfie of themselves, definitely stands out.“This is for creators who want to bring themselves into their content but don't want to have to shoot themselves,” said Elias Roman, vice president of product management at Google Labs.While creators may be interested in automating aspects of their video creation process, fully AI generated clips will likely be polarizing for audiences and may come across as inauthentic.