The big AI announcements from Google I/O 2023

(Image: Google)

Google CEO Sundar Pichai began the company's annual developer conference by acknowledging the popularity of AI, and what the company was planning to do with the technology.

Google has been fighting a hard battle to keep Microsoft and OpenAI away from its dominance in the search market, and at I/O 2023, Google came out swinging announcing a slew of new Generative AI features for its products.

The PaLM 2 foundation model

PaLM 2 is the upgraded version of Google's Pathways Language Model, or PaLM for short. Compared to the first iteration, PaLM 2 is heavily trained on multilingual text with support for more than 100 languages.

It can understand and generate nuanced text like poems or riddles across a variety of languages. It's also trained on scientific papers and web pages with mathematical expressions, giving it the ability to reason and solve math problems.

It can also analyze and generate code, having been trained in publicly available code sets in popular languages like Python and JavaScript.

The company announced that PaLM 2 will be at the heart of 25 Google products and services including the Bard AI chatbot.

Google's Bard AI no longer restricted by waitlist

Google has opened up its AI chatbot to users in over 180 countries worldwide. Previously, the chatbot was restricted to the United States and United Kingdom, gated behind a waitlist.

Now powered by PaLM 2, Bard will provide more in-depth answers, and can also help developers with programming and coding problems, including code generation and debugging.

The company is also making Bard's responses more visual, such as displaying charts or images to prompts where they are relevant.

Bard will support third-party extensions, similar to plugins from OpenAI, where it will allow developers to tap into Bard and integrate new functionalities.

Oh, and it finally has dark mode now.

Generative AI in other Google services

Gmail will soon help you draft mails, using a new feature called Help Me Write, that can generate drafts based on your prompts. It can even use information from earlier mails for context.

Help Me Write will be coming to Gmail, Google Docs, Google Sheets, and Google Slides as part of Google's Duet AI service updates, coming soon.

Generative AI updates in Search

Google announced new updates for Search that will make it easier for people to find what they are looking for, with detailed results and a new integrated search panel that will display AI-powered suggestions for links and other topics of interests based on the query.

It will even help you by coming up with relevant prompts for the AI based on the topic of interest. The company will also provide source links for all the information provided.

Besides this, to combat the ever-increasing problem of identifying fake AI-generated images, Google will introduce 'About This Image', that will display where the particular image originates from and how often it has appeared on Search.

The company said that all AI-generated images using its services will house metadata that indicate they have been AI-generated. Third-party creators will also have the ability to add relevant metadata.

Google Maps gets an AI infused Immersive View

Maps' Immersive View already uses Street View and AI to create an authentic representation of the world around you. Google is expanding Immersive View for Routes as well, giving you a detailed bird's eye view of the route you are planning to take, along with points of interest.

At launch, the feature will be available in 15 cities, including popular tourist destinations like Amsterdam, Berlin, Las Vegas, New York, Paris and Miami. So far, no Indian city seems to be on the list.

Generative AI features coming to Android

It isn't too much of a surprise that Generative AI will soon find its way into Android 14, starting with Magic Compose.

As the name suggests, the feature can help you draft replies using Google's Messaging app and even auto-suggest responses.

Android will also get a taste of Google's text-to-image model, which will allow users to generate AI wallpapers. Besides this, Android can convert your photos into 3D.

Project Tailwind

Tailwind is an AI service that will learn based on your notes. In short, it learns from your information and organizes it for you. It will also give you source links and citations as needed.

Tailwind is currently in waitlist at g.co/Labs.