Google I/O 2023’s Most Exciting AI Announcements

AI was the talk of the town at Google I/O 2023. Nearly all of the keynote speeches had something AI-related in it, causing the MIT Technology Review to say, “That wasn’t Google I/O — it was Google AI.”

AI is poised to change the digital landscape.

Love it or hate it, AI is transforming the way we work and play. Given the explosion of AI interest after the release of ChatGPT, Google is under pressure to incorporate it even more tightly into products used by billions of people. Google I/O 2023 gave us glimpses into Google’s aspirations, which range from helping developers find bugs more effectively, to letting Maps users visualize their entire route before they start, to helping people with physical disabilities control a mouse using only facial gestures.

So what are the most exciting AI announcements to come out of the conference? Here’s a breakdown of exciting Google I/O 2023 news for developers and for consumers.

5 Exciting AI Announcements for Google Developers

Developers can look forward to AI helping them code more effectively, learn new skills more quickly and even complete complex tasks with little to no experience. Here are some exciting Google I/O ’23 announcements that use AI to support coding:

1. Google’s Generative AI and PaLM 2

Google is leveraging the power of generative AI to improve their products, and they’re using PaLM 2 as the foundation model to power these experiences. To encourage developers to incorporate this technology into their own products, Google showcased several examples of ways to use the PaLM 2 API at Google I/O 2023, including:

• Enterprise Customer Chat Product: Got It AI uses the PaLM API to create summaries of customer conversations, helping businesses to better understand their customers’ needs and improve customer service.
• Interactive Gaming: Using the PaLM API, Play Labs built an agent that suggests ways to play and engage with friends, making the gaming experience more personalized and interactive.
• Intelligent Agents: Combining PaLM API with popular frameworks like LangChain gives developers the ability to build intelligent agents that use ReAct to navigate complex tasks. This could be applied in various fields, from customer service bots to personal digital assistants.
• Creating Descriptions using Large Language Models: Google created a playable trading card game called Flip I/O using PaLM 2. Google’s developers used generative AI to create millions of trading cards, each with their own unique images and descriptions. Game developers can use this tool to generate descriptions for characters and objects in games.

2. AI in Android Studio with Studio Bot

Studio Bot is an an AI-powered helper that Google integrated into Android Studio. Studio Bot makes developers more productive by recognizing natural language commands in a chat dialog. It assists with common developer tasks like generating code and debugging. One important part of this announcement is that Google assured developers that anything entered into Studio Bot will not be passed on to Google. There have been cases of abuse with ChatGPT, where sensitive information entered into the chat was publicly exposed, so verifying that Studio Bot will protect this information is a strong step toward alieving privacy concerns.

3. Project Gameface

Google is making Android devices more accessible with Project Gameface. It uses MediaPipe and AI models to allow users to control a mouse with their faces, turning a user’s own body into a game controller. This allows people like Lance Carr, who is quadriplegic, to control their devices without using their hands. Project Gameface will be open-sourced, allowing developers to make leaps in making their apps more accessible.

4. WebGPU and WebAssembly (Wasm) for AI-Ready Web

WebGPU is a new API designed to make the web AI-ready by unlocking the power of GPUs for rendering on the web. Additionally, Wasm now supports managed memory languages, giving apps written in languages like Dart and Flutter a huge boost in how fast they can run browser code compared to compiling to JavaScript. Developers can now also use Kotlin with Wasm.

5. Duet AI for Cloud Applications

With the introduction of Duet AI, a new generative AI-powered collaborator, developers of all skill levels can build cloud apps, call custom code models directly from Duet AI, and create new content like images and text, without any expertise in machine learning.

Now that you know how Google is aiming to improve development with AI, it’s time to look at how they’re planning on improving their consumer products.

Google I/O ’23’s AI Announcements for Consumers

Developers weren’t the only ones to get exciting announcements at Google I/O ’23. Everyday users are also getting AI advancements that will help them be more productive and have more fun with their Google-powered products. Here are some of the things that Google is promising to do with AI in the near future.

1. AI-Powered Search and Discovery

Microsoft has already integrated AI into their Edge browser and Bing search engine, and Google is following suit. They’ve removed the waiting list for their experimental AI chatbot, Bard, in 180 countries and territories (though not in the European Union, at least so far), and they’re integrating Bard into their search engine results pages. So what can you expect to see in the new Bard-fueled Google search engine?

• AI-powered snapshots: Snapshots will appear at the top of the Google search results pages. They’ll include relevant information for the searcher along with ways to get more information. Depending on the type of search, the snapshot might contain product images and prices, links to expert articles on the topic and corroborating information. The goal is to help consolidate information in complex searches.
• Conversational mode: With Bard powering search, you can ask more verbose and specific questions, getting more specific and creative suggestions that come closer to what you are looking for. You can even ask follow-up questions if you don’t get the results you want (and even some suggestions about what those follow-up questions might be)b

2. AI in Google Workspace

Google Workspace is getting several productivity upgrades thanks to AI. These include:

• Gmail’s “Help Me Write” feature, which will let you describe the email you want to write and let AI create it for you.
• Project Tailwind: A custom notebook for AI-assisted research and personalized AI models. You can ask Tailwind to create a study plan based off notes you stored in files you specify, for example.
• AI-assisted collaboration features in Workspace apps: For example, Sidekick provides intelligent suggestions and generates images based on the context of the document or task you’re working on.