Google I/O 2024 wrap: Google brings Gemini AI to Search, Android, and more

At its I/O 2024 Keynote, Google announced new AI models, tools, and deeper AI integration into its platforms such as Search and Android

Google I/O 2024 kicked off with a keynote address in which the US-based software giant focused mainly on deeper artificial intelligence (AI) integration into its core services such as Search, Workspace, Android operating system, and more. Google also announced its new Gemini 1.5 Pro AI model, which will be powering most of the new features within its services.

Gemini 1.5 Pro: Google said the new AI model is better at understanding the context behind the user’s prompt and has multimodal capabilities for better understanding text, images, videos and speech. The Gemini 1.5 model will also be available in a smaller scaled version called Gemini 1.5 Flash with 1 million tokens.

Google Project Astra: Google announced Project Astra, its “universal AI agent” with real-time spatial understanding. The AI agent has multimodal capabilities and it will power Gemini products starting later this year.

Generative Media models: Along with the next-generation versions of its Image generation model Imagen 3, Google launched Music AI sandbox for creating music tracks using AI and also a new generative video model called “Google Veo”.

Google Search: Google’s search engine is receiving more AI treatment with a custom Gemini AI model backing it. With the new update, users will be able to get AI-powered overviews in search results. The search engine also gets multi-step reasoning capabilities for breaking longer search queries to offer better search results. Google also said that it will bring a new Video Search feature later this year.

Gemini in Workspace: Google Workspace apps such as Gmail, Drive, Docs and more will get a new Gemini AI powered side panel for better accessibility. Additionally, Google launched a new Gemini AI teammate which will essentially work like an AI assistant on workspace apps.

AI in Android: Google’s mobile operating platform Android is getting deeper AI integration. The gesture driven Circle to Search on Android will now be able to solve complex mathematical problems while the Gemini AI assistant gets features like YouTube video summary, document summary and more.

Gemini app: The Gemini app for Android is getting a new Gemini Live feature which will allow the AI chatbot to understand and process live video for real time responses. Users can also create custom Gemini chatbots called “Gems” similar to custom GPTs on ChatGPT.

12:23 AM

Keynote ends

Sundar Pichai wraps-up the Keynote event

12:10 AM

Pali Gemma

• Google announces new vision language open model: Pali Gemma
• Gemma 2 coming in June

12:03 AM

Multimodality in Gemini Nano

Gemini Nano with multimodal capabilities is coming to Pixel devices later this year
Works on-device
Spam call detection
Understands what's on the display

screenshot

11:58 PM

AI on Android

AI integration
AI powered Search: Solve mathematical problems using Circle to Search
Gemini AI assistant: New design, YouTube Video understanding, Doccument summarisation

screenshot

11:44 PM

Gemini app

Gemini Live: Converstaional and live video understanding capability
Custom AI chatbot "Gems"
Trip planing capability with Gemini Advance
Bigger context window
2M tokens long context window later this year with Gemini Advance

screenshot

11:38 PM

Gemini AI Teammate

• AI-assistant for Workspace apps
• Has its own Google Account
• Can work in group spaces

screenshot

11:32 PM

Gemini for Workspace

Gemini powered side-panel in Workspace apps

Google Meet: Expanded language support

Gmail mobile app: New mobile card, Gemini 1.5 Pro powered summarising, Smart replies

screenshot

11:16 PM

Google Search with generative AI

• Powered by custom Gemini model 
• AI Overviews in Search
• Multi-step reasoning: Breaks bigger questions for better understanding
• Planning in Search: Creates multi-day plans in Search results
• AI-organised Search result page
• Video Search feature coming later this year

11:11 PM

Google Trillium

• New TPU
• Available to cloud customers later this year
  screenshot

11:06 PM

Google Veo

• Generative Video model
• 1080p videos from texts, images and video prompts
• Available in VideoFX for testing

screenshot

11:04 PM

Google Generative Media

Imagine 3: Google's latest image generation model
Remembers smaller details from user's prompt
Will be soon available to Enterprise customers and developers in Vertex AI

Music AI sandbox: AI-powered music generator
Made in colaboration with YouTube

10:58 PM

Google Project Astra

Google says it is an universal AI agent
Multimodal
Understands real-time data
Spatial understanding
Will come to Gemini products later this year

10:55 PM

Google launches Gemini 1.5 Flash

• • Small scale
  • 1M tokens
  • Available in AI Studio

  screenshot

10:49 PM

Multimodal AI model

Gemini 1.5 Pro is better at understanding context of the input  and offers personalised responses

10:46 PM

Google Workspaces

Gmail: With Gemini, summarise email from particular sender pointers

Google Meet: Provides highlights 

Gemini 1.5 Pro available in WorkSpace labs