Google IO 2024 - Key AI announcements

Post by Michael Rosario of InnovativeTeams.NET

I want to thank Google developer group program for enabling our team to attend Google IO. The experience has truly been inspiring. Our GDG team will work hard to bring the technology opportunities back to our developers in Florida. Beyond the amazing technology, I have enjoyed the gift of talking to other GDG organizers, technology trainers, and the teams behind our Google products. As a geek, it feels like Christmas. Thank you Google! In this post, I want to highlight some of the key announcements that we will unpack in the next months. To explore some of these topics early, check out the following links:

Welcome to the Gemini Era: Google AI Makes a Huge Leap Forward

This week at Google, the Google IO conference unveiled a wave of advancements in artificial intelligence (AI) designed to empower you in your everyday life, at work, and throughout your creative endeavors. It’s the start of the “Gemini era” for Google.

Introducing Gemini: A Powerful, Multi-Modal AI At the heart of these breakthroughs is Gemini, Google’s latest generative AI model. Unlike prior models, Gemini can analyze different data types beyond text, code, and images, and video to understand your needs better. This multi-model allows for incredible new features:

  • Ask Photos Anything: Use natural language to search your Google Photos library. Need to find a specific receipt or that time you went hiking? Just ask!

  • Large Context Window: With a 1 million token window, Gemini can understand complex requests and long sequences of information. This feels like a huge benefit over other LLM players. Google will be releasing a 2 million token context window very soon too.

  • Over 1.5 Million Developers usin Gemini: The Gemini 1.5 Pro model is now open to all developers, allowing them to build even more innovative applications.

AI Integration Across Google Products Gemini isn’t just a standalone tool. Google is integrating it across many of your favorite Google products:

  • Workspace Revolution: Gemini can summarize emails and meetings in Google Workspace, helping you stay on top of your workload.

  • Note-Taking with a Twist: The new Google Notebook LM integrates with Gemini. This tool enables you to generate audio summaries from your notes or even create music based on your ideas. This tool enabled users to make a personalized podcast-style summary of recent notes that help with recall and study engagement.

  • AI Agents That Think Ahead: Google is developing AI agents that can reason and plan multiple steps ahead, working alongside you to tackle complex tasks like planning a move or returning an item. I really liked their demo of practical meal planning using Gemini AI features in Google Search. Gemini enabled Google search features will be rolling out today!

  • Pushing the Boundaries of AI Research: Google DeepMind team continues to break new ground. Google discussed advancements in general AI research, including Alpha Fold 3D’s ability to solve protein folding problems robustly. The process of protein folding and finding real 3D structures to proteins is essential in the development of new drugs and curing diseases. Google also introduced Gemini 1.5 Flash, a lighter-weight version of the Pro model.

Project Astra

Project Astra was a mind-blowing research effort from the DeepMind team showing some of the baby steps toward GenAI. In the concept demo, the agent worked in real-time interpreting a video stream. In this mode, the system could identify objects in the room, review and reflect on code, and help the user find lost objects in the space. During the demo, the user talked in a very natural tone to the agent and showed impressive short-term memory recall.

Generative AI Takes Center Stage We’re bringing the power of generative AI to a wider audience than ever before:

  • Image and Music Creation: Labs.google is now home to ImageFX, a tool for manipulating images, and the Music AI Sandbox, where you can experiment with generating your own music.
  • Video Made Easy: Veo allows you to generate video content with VideoFX and tools like VideoPoet.
  • Search Gets Smarter with Generative AI Search is getting a major upgrade thanks to generative AI. We’re introducing new features to AI Overviews, including multi-step reasoning and the ability to break down complex questions into smaller, more manageable steps.
  • Gemini in the Workplace: Your AI Teammate Imagine having an AI assistant that can help you write, visualize data, and manage your workload. The Gemini side panel in Workspace aims to do just that, boosting productivity by 30%. In future editions of Google workspace, you can create a “project assistant” agent. This project again goes way beyond simple LLM summarization of a single document. Like a human assistant, the agent can have an awareness of key project artifacts, project schedule, budget, and scope. The agent can collaborate with other human actors in team chat. You can ask the project agent can support project leaders with time-consuming administrivia.

Excited to share insights from GoogleIO Day 2 tomorrow!