Google I/O 2024, May 14, 2024, is Google’s annual event that showcases its latest product launches and updates. This year, the focus is on Generative AI.
Announcements in Google I/O event May 14, 2024
Gemini 1.5 Model
All Gemini Advanced subscribers have acces to Gemini 1.5 model.
Context Window Expanded to 1 million tokens (750,000 words) and will soon expand to 2 million tokens (1.5 million words).
This significantly increases the amount of input and output data the model can handle, enhancing its capability for complex tasks.
Ask Your Photos Feature
This feature leverages AI to make personal photo libraries more searchable and useful. Examples: Users can ask questions like “What’s my license plate number?” or “When did Lucy learn to swim?” and the AI will search through photos to find the relevant information.
Gemini in Gmail
Summarizes and extracts information from emails, such as announcements from a child’s school. Organizing and summarizing email content.
Notebook LM
Combines documents and audio notes into a podcast-like format, allowing users to interject with questions during playback.
This is innovative approach to note-taking and information review, making it more interactive and engaging.
Working towards AI Agents
An agent performs multi-step tasks like returning shoes by interacting with various services and information. This represents a significant step towards more autonomous AI-driven task management.
Project Astra
Real-time AI agent using a camera to interpret and answer questions about the environment. This Demonstrates advanced real-time visual processing, offering potential applications in various fields.
Imagine 3
This is Google’s image generation platform, now capable of embedding text within images. It competes with similar tools like DALL-E, enhancing creative capabilities.
Music Effects
A generative music tool that provides creative musical options.
Veo
Veo is Video Generation Model that generates videos in 1080p, capable of longer than 60 seconds.
It competes with Sora, offering promising text-to-video capabilities with an open waitlist for users.
Google Search AI Integration
Introducing multi-step reasoning in search, providing detailed answers to complex queries. Enhanced AI-driven search with step-by-step reasoning. Aims to make search more intuitive and comprehensive. You give it multiple steps to answer and it provides answers to all questions. E.g. searching for a shoe store in Tallinn Mustamäe district which sells men’s dress shoes and has currently discount sale. This is new approach to search engine use, making searches more conversational and detailed.
Real-Time Captioning
Provides real-time captions and summarization across multiple emails.
Workflow Automation
Allows creation of repeatable workflows using Gemini.
Useful for automating repetitive tasks, increasing efficiency.
Gems
Pre-trained chat models with system prompts for consistent outputs. Similar to OpenAI’s GPTs, providing tailored conversational experiences.
Scam Detection
AI on Android phones that detects potential scams during phone calls. Adds a layer of security for users, which could be highly beneficial if adopted widely.
Open-Source Models
PaliGemma – Multmodal model that is open source. Can see images. Anybody can build upon that.
Gemma2 – Open source model, 27B parameters.
References
https://io.google/2024/
https://www.tomsguide.com/news/live/google-io-2024-keynote