Google announces Project Astra as "AI agent for everyday life"

At Google's I/O 2024, Demis Hassabis, CEO of deepmind gave his first ever appearances on stage about Google’s "Universal AI agent for everyday life” or in short ‘Project Astra’. Project Astra would be loaded with features like object recognition made up of AI agents as they call it. And, in this article, we will discuss further about Google’s Project Astra in more detail. 

Google's Project Astra Overview

Google's new AI progress, particularly with Gemini models, is a big deal in tech. Gemini 1.5 Flash, which you can try out now, makes those large language models faster and cheaper. Plus, with improvements to Gemini 1.5 Pro, like a bigger context window and better responses, tasks such as translation and coding become even smoother.

What can it do? 

The Project Astra can describe things, give you information, and help you in everyday life as well. Through the Project Astra, you can simply show what's in front of you, and it will answer pretty accurately. It can also help you with your assignments or even with your coding.

Project Astra is a prototype from @GoogleDeepMind exploring how a universal AI agent can be truly helpful in everyday life. Watch our prototype in action in two parts, each captured in a single take, in real time ↓ #GoogleIO pic.twitter.com/uMEjIJpsjO

Advertisement

— Google (@Google)

What Else from the I/O?

Expanding Gemini's Reach

Gemini Nano's evolution to incorporate image understanding brings multimodal capabilities to smartphone applications, underscoring Google's commitment to versatility and accessibility. Additionally, the introduction of Gemma 2 and PaliGemma models, optimized for TPUs and GPUs respectively, showcases Google's dedication to optimizing AI performance across various hardware platforms.

Google Search Gets Smarter

Google's Gemini Nano now understands images, making smartphone apps even smarter. This shows Google's focus on making AI useful for everyone. Also, the new Gemma 2 and PaliGemma models, optimized for TPUs and GPUs (computer chips), show how Google is making AI work better on all sorts of devices.

Veo, Imagen 3 and SynthID

Google unveils Veo, a groundbreaking model that turns text into high-quality 1080p videos. Veo can understand both regular language and cinematic terms, letting users create videos that match their style and vision. Though it's only available to a select few for now, Veo plans to open up to more creators with a waitlist. Google's latest innovation, Imagen 3, takes text and turns it into images with better natural language skills, promising top-notch image quality. The new SynthID tech now also protects text and video content, ensuring it's authentic and safe, especially in videos made by Veo.

• Also, read
  • Google is exploring options to use Chromium OS on Android devices
  • Report claims SMIC has finally developed its 5nm node with EUV
  • Xiaomi Smart Camera C700 launched with 4K recording and HyperOS

Ask Photos

Google Photos introduces Ask Photos, enabling conversational image search and automatic generation of image highlights. Meanwhile, Gemini's integration into Android phones as the default AI assistant brings multimodal support and enhanced accessibility features to users' fingertips.

Google's Project Astra: What’s next?

Google has big plans for Gemini, aiming to make it even better with features like Gemini Live for live voice chats and Gems for personalized AI help. Plus, integrating Gemini into Google Workspace means smarter help with all kinds of work tasks, showing Google's dedication to making AI boost productivity for everyone.

• Meanwhile, watch our review of Honor Magic V2