Google I/O 2024 Announcements - Focus on Gemini and AI

AI advancements

• Introduced Gemini 1.5: a faster, lighter-weight AI model with a 1 million token context window

• Announced Trillium, the most performant TPU to date with 4.7x faster compute performance

• Unveiled Imagen 3, capable of generating high-quality photorealistic images with rich details

• Showcased Veo, a new video generation model that creates cinematic videos in various styles
  

New features in Gemini App

• Gemini 1.5 Pro available for Advanced subscribers, offering the largest context window of any commercial chatbot

• Upload files directly into Gemini Advanced for analysis and data visualization

• New trip planning feature creates custom itineraries

• Gemini Live enables natural spoken conversations with Gemini through voice commands
  

Search Updates

• Implemented a new Gemini model for Google Search, combining advanced reasoning and multimodality

• Rolled out AI Overviews in Search to provide summaries of search results

• Introduced multi-step reasoning capabilities for complex search queries

• Search will offer planning features for meals, trips, and more
  

Google I/O 2024 Announcements focusing on Gemini and AI.

Integration with Workspace and Photos

• Gemini 1.5 Pro available in the side panel of Gmail, Docs, Drive, Slides, and Sheets

• New features in Gmail include email summaries, contextual Smart Reply, and Q&A

• Ask Photos, a new feature in Google Photos, uses Gemini to search for memories and generate captions
  

Android advancements

• Pixel phones to feature multimodal capabilities with Gemini Nano

• Improved accessibility features with Talkback powered by Gemini Nano

• New on-device scam protection using Gemini Nano

• Integration of Circle to Search with various apps for knowledge acquisition

• Introduced Android 15 with Theft Detection Lock, Private Space, and improved security features
  

Developments for Developers

• Gemini API Developer Competition with a unique prize

• Unveiled PaliGemma, a vision-language open model for visual Q&A and image captioning

• Previewed Gemma 2, a larger and more powerful language model

• Integration of Gemini models with various development tools

• Introduced new features in the Gemini API for smoother workflows
  

Responsible AI

• Enhanced red teaming with AI-Assisted Red Teaming

• Expanded SynthID to text and video formats

• Open-sourced SynthID text watermarking

• Introduced LearnLM, a family of models focused on learning

• Collaboration with institutions to refine LearnLM

• Developed an online course on using generative AI in education
  

Other announcements

• Launched Illuminate, a tool that generates conversations to explain research papers

Source: Google Blog - https://blog.google/technology/ai/google-io-2024-100-announcements/