Google I/O 2024 wrapped up earlier on Tuesday with a bang. Google made several announcements with a heavy focus on Gemini AI models as well as Android 15 updates. There was no hardware update at the Google I/O 2024 unlike usual since Pixel 8a was announced last week. Cut to now, let’s have a recap of all the major announcements made at Google I/O 2024 that you should be on the lookout for.
Google’s Gemini AI Gets Faster
One of the highlights of Google I/O 2024 was the Gemini AI. Google just introduced a new AI model called Gemini 1.5 Flash, an equally powerful multimodal model as Gemini 1.5 Pro but optimized for high-frequency, narrow, and low-latency tasks in mind.
Additionally, Google has improved upon the existing Gemini 1.5 model for better translation, reasoning, and coding. The Gemini 1.5 Pro too has doubled the amount of information it can take in (called context window) from 1 million to 2 million tokens.
Along with that, Google Gemini 1.5 Pro and Flash is now available across 200 countries while Gemma 2 will follow the wagon in June 2024.
Google Lens Now Works on Videos
Google Lens has shown its caliber to operate efficiently on images as you can search for anything. With that being said, Google is taking it a step further with the ability to search on videos. Simply shoot a video and ask Google Lens about the video such as ‘Why the water boils at this temperature’ and Google Lens should be able to interpret and answer. Of course, Google’s AI will work under the hood to pull relevant results from the internet.
Google AI call monitoring to detect spams
One of the feature that we got to see at Google I/O 2024 is about call monitoring to detect spams. Apparently, Google Gemini AI will be able to listen to all your calls and prompt you in case if it senses fraudulent activities. A demo that went viral at IO 2024 showed a test call where the person on the other side asked to transfer money to a different account. Google promptly sent a warning that the call looks a scam since banks never ask you to transfer money directly.
The feature is currently under a lot of scrutiny among fans given its ability to eavesdrop on all
your calls raising privacy concerns for many.
Google AI for Learning: LearnLM Announced at Google I/O 2024
Google I/O 2024 saw yet another family of generative AI models for learning dubbed as LearnLM. Made in collaboration with Google Research and DeepMind AI, the models are built on top of Google Gemini and made to conversationally tutor students when using Google products. This includes Google Classroom, Google Search, YouTube, Gemini apps, and beyond.
Introduced as a pilot program on Google Chrome, the AI would make it easier for educators to improve their lesson planning. It will also help them uncover new ideas, content, and materials targeting the specific needs of the students.
Gemini Nano Gets Multi-modality
Gemini Nano is Google’s smallest mobile large language mode (LLM) that just got a boost in terms of Multi-modality. The feature gives AI the ability to turn any input into any output as per Google CEO Sundar Pichai. Multi-modality helps Gemini Nano pull information from social media videos, web audio, photos, live videos, and texts and beyond to summarize the input and answer your questions.
Circle to Search Supports Step-by-Step Answers
Samsung popularized the ‘Circle to Search’ feature when it made its debut on the Galaxy S24 series. Soon, Google Pixel phones got these AI capabilities too thanks to Google Lens. Cut to now, Google is extending Circle to Search to support step-by-step answers. It is helpful for students as now, they can circle any math or other problems and get detailed step-by-step answers on how to solve them. This is a part of Search Labs beta and requires more work, especially with LearnLM before it can replace alternatives.
Google Adds a Suite of Creativity Tools
Google I/O 2024 was packed with all things AI. Google CEO Sundar Pichai used Gemini to count the number of times ‘AI’ was mentioned, and it was exactly 120 times. With that being said, Google has upped its game in terms of AI image generators with Imagen3 capable of producing photorealistic photos. It is currently the best model yet (from Google) for text-to-image generation. It will be available in ImageFX via labs. Google to users while developers should get it by June through Vertex AI.
That’s not all though. Google has added VideoFX, an AI-based video generative model based on Google DeepMind’s Veo video generator. It lets you generate 1080p videos based on text prompts. We aren’t done yet. There’s a new DJ Mode in MusicFX — apparently, an AI music generator, that gives you samples and song loops based on prompts.
Google Play Store upping protection against fraudulent apps
Google Play Store keeps an eye on apps to steer you away from bad ones out there. With the AI boost Google is pushing through, Google Play Store is getting an on-device AI to find apps committing any kind of fraud or phishing activities. The data will be relayed to Google for review. It also warns users about such apps providing an additional layer against getting phished or scammed.
