On the finish of I/O, Google’s annual developer convention on the Shoreline Amphitheater in Mountain View, Google CEO Sundar Pichai revealed that the corporate had stated “AI” 121 instances. That, basically, was the crux of Google’s two-hour keynote — stuffing AI into each Google app and repair utilized by greater than two billion individuals around the globe. Listed here are all the key updates that Google introduced on the occasion.
Gemini 1.5 Flash and updates to Gemini 1.5 Professional
Google introduced a model new AI mannequin referred to as Gemini 1.5 Flash, which it says is optimised for velocity and effectivity. Flash sits between Gemini 1.5 Professional and Gemini 1.5 Nano, which its the corporate’s smallest mannequin that runs regionally on system. Google stated that it created Flash as a result of builders needed a lighter and cheaper mannequin than Gemini Professional to construct AI-powered apps and companies whereas holding a number of the issues like a protracted context window of 1 million tokens that differentiates Gemini Professional from competing fashions. Later this yr, Google will double Gemini’s context window to 2 million tokens, which signifies that it will likely be in a position to course of two hours of video, 22 hours of audio, greater than 60,000 strains of code or greater than 1.4 million phrases on the similar time.
Mission Astra
Google confirmed off Mission Astra, an early model of a common assistant powered by AI that Google’s DeepMind CEO Demis Hassabis stated was Google’s model of an AI agent “that may be useful in on a regular basis life.”
In a video that Google says was shot in a single take, an Astra consumer strikes round Google’s London workplace holding up their cellphone and pointing the digicam at numerous issues — a speaker, some code on a whiteboard, and out a window — and has a pure dialog with the app about what it appears. In one of many video’s most spectacular moments, the appropriately tells the consumer the place she left her glasses earlier than with out the consumer ever having introduced up the glasses.
The video ends with a twist — when the consumer finds and wears the lacking glasses, we study that they’ve an onboard digicam system and are able to utilizing Mission Astra to seamlessly stick with it a dialog with the consumer, maybe indicating that Google is perhaps engaged on a competitor to Meta’s Ray Ban sensible glasses.
Ask Google Pictures
Google Pictures was already clever when it got here to looking for particular photographs or movies, however with AI, Google is taking issues to the following stage. In case you’re a Google One subscriber within the US, it is possible for you to to ask Google Pictures a posh query like “present me the very best photograph from every nationwide park I’ve visited” when the function rolls out over the following few months. Google Pictures will use GPS data in addition to its personal judgement of what’s “finest” to current you with choices. You too can ask Google Pictures to generate captions to publish the photographs to social media.
Veo and Imagen 3
Google’s new AI-powered media creation engines are referred to as Veo and Imagen 3. Veo is Google’s reply to OpenAI’s Sora. It could possibly produce “high-quality” 1080p movies that may final “past a minute”, Google stated, and may perceive cinematic ideas like a timelapse.
Imagen 3, in the meantime, is a text-to-image generator that Google claims handles textual content higher than its earlier model, Imagen 2. The result’s the corporate’s highest high quality” text-to-image mannequin with “unbelievable stage of element” for “photorealistic, lifelike photographs” and fewer artifacts — basically pitting it towards OpenAI’s DALLE-3.
Huge updates to Google Search
Google is making massive modifications to how Search essentially works. Many of the updates introduced at this time like the power to ask actually advanced questions (“Discover the very best yoga or pilates studios in Boston and present particulars on their intro gives and strolling time from Beacon Hill.”) and utilizing Search to plan meals and holidays gained’t be accessible until you choose in to Search Labs, the corporate’s platform that lets individuals check out experimental options.
However an enormous new function that Google is asking AI Overviews and which the corporate has been testing for a yr now, is lastly rolling out to hundreds of thousands of individuals within the US. Google Search will now current AI-generated solutions on prime of the outcomes by default, and the corporate says that it’ll deliver the function to greater than a billion customers around the globe by the tip of the yr.
Gemini on Android
Google is integrating Gemini instantly into Android. When Android 15 releases later this yr, Gemini will pay attention to the app, picture or video that you just’re operating, and also you’ll be capable of pull it up as an overlay and ask it context-specific questions. The place does that go away Google Assistant that already does this? Who is aware of! Google didn’t deliver it up in any respect throughout at this time’s keynote.
There have been a bunch of different updates too. Google stated it will add digital watermarks to AI-generated video and textual content, make Gemini accessible within the facet panel in Gmail and Docs, energy a digital AI teammate in Workspace, eavesdrop on cellphone calls and detect for those who’re being scammed in actual time, and much more.
Make amends for all of the information from Google I/O 2024 proper right here!